Merge branch 'ta/string-list-init'
[git] / diff.c
1 /*
2  * Copyright (C) 2005 Junio C Hamano
3  */
4 #include "cache.h"
5 #include "quote.h"
6 #include "diff.h"
7 #include "diffcore.h"
8 #include "delta.h"
9 #include "xdiff-interface.h"
10 #include "color.h"
11 #include "attr.h"
12 #include "run-command.h"
13 #include "utf8.h"
14 #include "userdiff.h"
15 #include "sigchain.h"
16 #include "submodule.h"
17 #include "ll-merge.h"
18 #include "string-list.h"
19 #include "argv-array.h"
20
21 #ifdef NO_FAST_WORKING_DIRECTORY
22 #define FAST_WORKING_DIRECTORY 0
23 #else
24 #define FAST_WORKING_DIRECTORY 1
25 #endif
26
27 static int diff_detect_rename_default;
28 static int diff_rename_limit_default = 400;
29 static int diff_suppress_blank_empty;
30 static int diff_use_color_default = -1;
31 static int diff_context_default = 3;
32 static const char *diff_word_regex_cfg;
33 static const char *external_diff_cmd_cfg;
34 static const char *diff_order_file_cfg;
35 int diff_auto_refresh_index = 1;
36 static int diff_mnemonic_prefix;
37 static int diff_no_prefix;
38 static int diff_stat_graph_width;
39 static int diff_dirstat_permille_default = 30;
40 static struct diff_options default_diff_options;
41 static long diff_algorithm;
42
43 static char diff_colors[][COLOR_MAXLEN] = {
44         GIT_COLOR_RESET,
45         GIT_COLOR_NORMAL,       /* PLAIN */
46         GIT_COLOR_BOLD,         /* METAINFO */
47         GIT_COLOR_CYAN,         /* FRAGINFO */
48         GIT_COLOR_RED,          /* OLD */
49         GIT_COLOR_GREEN,        /* NEW */
50         GIT_COLOR_YELLOW,       /* COMMIT */
51         GIT_COLOR_BG_RED,       /* WHITESPACE */
52         GIT_COLOR_NORMAL,       /* FUNCINFO */
53 };
54
55 static int parse_diff_color_slot(const char *var)
56 {
57         if (!strcasecmp(var, "plain"))
58                 return DIFF_PLAIN;
59         if (!strcasecmp(var, "meta"))
60                 return DIFF_METAINFO;
61         if (!strcasecmp(var, "frag"))
62                 return DIFF_FRAGINFO;
63         if (!strcasecmp(var, "old"))
64                 return DIFF_FILE_OLD;
65         if (!strcasecmp(var, "new"))
66                 return DIFF_FILE_NEW;
67         if (!strcasecmp(var, "commit"))
68                 return DIFF_COMMIT;
69         if (!strcasecmp(var, "whitespace"))
70                 return DIFF_WHITESPACE;
71         if (!strcasecmp(var, "func"))
72                 return DIFF_FUNCINFO;
73         return -1;
74 }
75
76 static int parse_dirstat_params(struct diff_options *options, const char *params_string,
77                                 struct strbuf *errmsg)
78 {
79         char *params_copy = xstrdup(params_string);
80         struct string_list params = STRING_LIST_INIT_NODUP;
81         int ret = 0;
82         int i;
83
84         if (*params_copy)
85                 string_list_split_in_place(&params, params_copy, ',', -1);
86         for (i = 0; i < params.nr; i++) {
87                 const char *p = params.items[i].string;
88                 if (!strcmp(p, "changes")) {
89                         DIFF_OPT_CLR(options, DIRSTAT_BY_LINE);
90                         DIFF_OPT_CLR(options, DIRSTAT_BY_FILE);
91                 } else if (!strcmp(p, "lines")) {
92                         DIFF_OPT_SET(options, DIRSTAT_BY_LINE);
93                         DIFF_OPT_CLR(options, DIRSTAT_BY_FILE);
94                 } else if (!strcmp(p, "files")) {
95                         DIFF_OPT_CLR(options, DIRSTAT_BY_LINE);
96                         DIFF_OPT_SET(options, DIRSTAT_BY_FILE);
97                 } else if (!strcmp(p, "noncumulative")) {
98                         DIFF_OPT_CLR(options, DIRSTAT_CUMULATIVE);
99                 } else if (!strcmp(p, "cumulative")) {
100                         DIFF_OPT_SET(options, DIRSTAT_CUMULATIVE);
101                 } else if (isdigit(*p)) {
102                         char *end;
103                         int permille = strtoul(p, &end, 10) * 10;
104                         if (*end == '.' && isdigit(*++end)) {
105                                 /* only use first digit */
106                                 permille += *end - '0';
107                                 /* .. and ignore any further digits */
108                                 while (isdigit(*++end))
109                                         ; /* nothing */
110                         }
111                         if (!*end)
112                                 options->dirstat_permille = permille;
113                         else {
114                                 strbuf_addf(errmsg, _("  Failed to parse dirstat cut-off percentage '%s'\n"),
115                                             p);
116                                 ret++;
117                         }
118                 } else {
119                         strbuf_addf(errmsg, _("  Unknown dirstat parameter '%s'\n"), p);
120                         ret++;
121                 }
122
123         }
124         string_list_clear(&params, 0);
125         free(params_copy);
126         return ret;
127 }
128
129 static int parse_submodule_params(struct diff_options *options, const char *value)
130 {
131         if (!strcmp(value, "log"))
132                 DIFF_OPT_SET(options, SUBMODULE_LOG);
133         else if (!strcmp(value, "short"))
134                 DIFF_OPT_CLR(options, SUBMODULE_LOG);
135         else
136                 return -1;
137         return 0;
138 }
139
140 static int git_config_rename(const char *var, const char *value)
141 {
142         if (!value)
143                 return DIFF_DETECT_RENAME;
144         if (!strcasecmp(value, "copies") || !strcasecmp(value, "copy"))
145                 return  DIFF_DETECT_COPY;
146         return git_config_bool(var,value) ? DIFF_DETECT_RENAME : 0;
147 }
148
149 long parse_algorithm_value(const char *value)
150 {
151         if (!value)
152                 return -1;
153         else if (!strcasecmp(value, "myers") || !strcasecmp(value, "default"))
154                 return 0;
155         else if (!strcasecmp(value, "minimal"))
156                 return XDF_NEED_MINIMAL;
157         else if (!strcasecmp(value, "patience"))
158                 return XDF_PATIENCE_DIFF;
159         else if (!strcasecmp(value, "histogram"))
160                 return XDF_HISTOGRAM_DIFF;
161         return -1;
162 }
163
164 /*
165  * These are to give UI layer defaults.
166  * The core-level commands such as git-diff-files should
167  * never be affected by the setting of diff.renames
168  * the user happens to have in the configuration file.
169  */
170 int git_diff_ui_config(const char *var, const char *value, void *cb)
171 {
172         if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
173                 diff_use_color_default = git_config_colorbool(var, value);
174                 return 0;
175         }
176         if (!strcmp(var, "diff.context")) {
177                 diff_context_default = git_config_int(var, value);
178                 if (diff_context_default < 0)
179                         return -1;
180                 return 0;
181         }
182         if (!strcmp(var, "diff.renames")) {
183                 diff_detect_rename_default = git_config_rename(var, value);
184                 return 0;
185         }
186         if (!strcmp(var, "diff.autorefreshindex")) {
187                 diff_auto_refresh_index = git_config_bool(var, value);
188                 return 0;
189         }
190         if (!strcmp(var, "diff.mnemonicprefix")) {
191                 diff_mnemonic_prefix = git_config_bool(var, value);
192                 return 0;
193         }
194         if (!strcmp(var, "diff.noprefix")) {
195                 diff_no_prefix = git_config_bool(var, value);
196                 return 0;
197         }
198         if (!strcmp(var, "diff.statgraphwidth")) {
199                 diff_stat_graph_width = git_config_int(var, value);
200                 return 0;
201         }
202         if (!strcmp(var, "diff.external"))
203                 return git_config_string(&external_diff_cmd_cfg, var, value);
204         if (!strcmp(var, "diff.wordregex"))
205                 return git_config_string(&diff_word_regex_cfg, var, value);
206         if (!strcmp(var, "diff.orderfile"))
207                 return git_config_pathname(&diff_order_file_cfg, var, value);
208
209         if (!strcmp(var, "diff.ignoresubmodules"))
210                 handle_ignore_submodules_arg(&default_diff_options, value);
211
212         if (!strcmp(var, "diff.submodule")) {
213                 if (parse_submodule_params(&default_diff_options, value))
214                         warning(_("Unknown value for 'diff.submodule' config variable: '%s'"),
215                                 value);
216                 return 0;
217         }
218
219         if (!strcmp(var, "diff.algorithm")) {
220                 diff_algorithm = parse_algorithm_value(value);
221                 if (diff_algorithm < 0)
222                         return -1;
223                 return 0;
224         }
225
226         if (git_color_config(var, value, cb) < 0)
227                 return -1;
228
229         return git_diff_basic_config(var, value, cb);
230 }
231
232 int git_diff_basic_config(const char *var, const char *value, void *cb)
233 {
234         const char *name;
235
236         if (!strcmp(var, "diff.renamelimit")) {
237                 diff_rename_limit_default = git_config_int(var, value);
238                 return 0;
239         }
240
241         if (userdiff_config(var, value) < 0)
242                 return -1;
243
244         if (skip_prefix(var, "diff.color.", &name) ||
245             skip_prefix(var, "color.diff.", &name)) {
246                 int slot = parse_diff_color_slot(name);
247                 if (slot < 0)
248                         return 0;
249                 if (!value)
250                         return config_error_nonbool(var);
251                 color_parse(value, var, diff_colors[slot]);
252                 return 0;
253         }
254
255         /* like GNU diff's --suppress-blank-empty option  */
256         if (!strcmp(var, "diff.suppressblankempty") ||
257                         /* for backwards compatibility */
258                         !strcmp(var, "diff.suppress-blank-empty")) {
259                 diff_suppress_blank_empty = git_config_bool(var, value);
260                 return 0;
261         }
262
263         if (!strcmp(var, "diff.dirstat")) {
264                 struct strbuf errmsg = STRBUF_INIT;
265                 default_diff_options.dirstat_permille = diff_dirstat_permille_default;
266                 if (parse_dirstat_params(&default_diff_options, value, &errmsg))
267                         warning(_("Found errors in 'diff.dirstat' config variable:\n%s"),
268                                 errmsg.buf);
269                 strbuf_release(&errmsg);
270                 diff_dirstat_permille_default = default_diff_options.dirstat_permille;
271                 return 0;
272         }
273
274         if (starts_with(var, "submodule."))
275                 return parse_submodule_config_option(var, value);
276
277         return git_default_config(var, value, cb);
278 }
279
280 static char *quote_two(const char *one, const char *two)
281 {
282         int need_one = quote_c_style(one, NULL, NULL, 1);
283         int need_two = quote_c_style(two, NULL, NULL, 1);
284         struct strbuf res = STRBUF_INIT;
285
286         if (need_one + need_two) {
287                 strbuf_addch(&res, '"');
288                 quote_c_style(one, &res, NULL, 1);
289                 quote_c_style(two, &res, NULL, 1);
290                 strbuf_addch(&res, '"');
291         } else {
292                 strbuf_addstr(&res, one);
293                 strbuf_addstr(&res, two);
294         }
295         return strbuf_detach(&res, NULL);
296 }
297
298 static const char *external_diff(void)
299 {
300         static const char *external_diff_cmd = NULL;
301         static int done_preparing = 0;
302
303         if (done_preparing)
304                 return external_diff_cmd;
305         external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
306         if (!external_diff_cmd)
307                 external_diff_cmd = external_diff_cmd_cfg;
308         done_preparing = 1;
309         return external_diff_cmd;
310 }
311
312 static struct diff_tempfile {
313         const char *name; /* filename external diff should read from */
314         char hex[41];
315         char mode[10];
316         char tmp_path[PATH_MAX];
317 } diff_temp[2];
318
319 typedef unsigned long (*sane_truncate_fn)(char *line, unsigned long len);
320
321 struct emit_callback {
322         int color_diff;
323         unsigned ws_rule;
324         int blank_at_eof_in_preimage;
325         int blank_at_eof_in_postimage;
326         int lno_in_preimage;
327         int lno_in_postimage;
328         sane_truncate_fn truncate;
329         const char **label_path;
330         struct diff_words_data *diff_words;
331         struct diff_options *opt;
332         int *found_changesp;
333         struct strbuf *header;
334 };
335
336 static int count_lines(const char *data, int size)
337 {
338         int count, ch, completely_empty = 1, nl_just_seen = 0;
339         count = 0;
340         while (0 < size--) {
341                 ch = *data++;
342                 if (ch == '\n') {
343                         count++;
344                         nl_just_seen = 1;
345                         completely_empty = 0;
346                 }
347                 else {
348                         nl_just_seen = 0;
349                         completely_empty = 0;
350                 }
351         }
352         if (completely_empty)
353                 return 0;
354         if (!nl_just_seen)
355                 count++; /* no trailing newline */
356         return count;
357 }
358
359 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
360 {
361         if (!DIFF_FILE_VALID(one)) {
362                 mf->ptr = (char *)""; /* does not matter */
363                 mf->size = 0;
364                 return 0;
365         }
366         else if (diff_populate_filespec(one, 0))
367                 return -1;
368
369         mf->ptr = one->data;
370         mf->size = one->size;
371         return 0;
372 }
373
374 /* like fill_mmfile, but only for size, so we can avoid retrieving blob */
375 static unsigned long diff_filespec_size(struct diff_filespec *one)
376 {
377         if (!DIFF_FILE_VALID(one))
378                 return 0;
379         diff_populate_filespec(one, 1);
380         return one->size;
381 }
382
383 static int count_trailing_blank(mmfile_t *mf, unsigned ws_rule)
384 {
385         char *ptr = mf->ptr;
386         long size = mf->size;
387         int cnt = 0;
388
389         if (!size)
390                 return cnt;
391         ptr += size - 1; /* pointing at the very end */
392         if (*ptr != '\n')
393                 ; /* incomplete line */
394         else
395                 ptr--; /* skip the last LF */
396         while (mf->ptr < ptr) {
397                 char *prev_eol;
398                 for (prev_eol = ptr; mf->ptr <= prev_eol; prev_eol--)
399                         if (*prev_eol == '\n')
400                                 break;
401                 if (!ws_blank_line(prev_eol + 1, ptr - prev_eol, ws_rule))
402                         break;
403                 cnt++;
404                 ptr = prev_eol - 1;
405         }
406         return cnt;
407 }
408
409 static void check_blank_at_eof(mmfile_t *mf1, mmfile_t *mf2,
410                                struct emit_callback *ecbdata)
411 {
412         int l1, l2, at;
413         unsigned ws_rule = ecbdata->ws_rule;
414         l1 = count_trailing_blank(mf1, ws_rule);
415         l2 = count_trailing_blank(mf2, ws_rule);
416         if (l2 <= l1) {
417                 ecbdata->blank_at_eof_in_preimage = 0;
418                 ecbdata->blank_at_eof_in_postimage = 0;
419                 return;
420         }
421         at = count_lines(mf1->ptr, mf1->size);
422         ecbdata->blank_at_eof_in_preimage = (at - l1) + 1;
423
424         at = count_lines(mf2->ptr, mf2->size);
425         ecbdata->blank_at_eof_in_postimage = (at - l2) + 1;
426 }
427
428 static void emit_line_0(struct diff_options *o, const char *set, const char *reset,
429                         int first, const char *line, int len)
430 {
431         int has_trailing_newline, has_trailing_carriage_return;
432         int nofirst;
433         FILE *file = o->file;
434
435         fputs(diff_line_prefix(o), file);
436
437         if (len == 0) {
438                 has_trailing_newline = (first == '\n');
439                 has_trailing_carriage_return = (!has_trailing_newline &&
440                                                 (first == '\r'));
441                 nofirst = has_trailing_newline || has_trailing_carriage_return;
442         } else {
443                 has_trailing_newline = (len > 0 && line[len-1] == '\n');
444                 if (has_trailing_newline)
445                         len--;
446                 has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
447                 if (has_trailing_carriage_return)
448                         len--;
449                 nofirst = 0;
450         }
451
452         if (len || !nofirst) {
453                 fputs(set, file);
454                 if (!nofirst)
455                         fputc(first, file);
456                 fwrite(line, len, 1, file);
457                 fputs(reset, file);
458         }
459         if (has_trailing_carriage_return)
460                 fputc('\r', file);
461         if (has_trailing_newline)
462                 fputc('\n', file);
463 }
464
465 static void emit_line(struct diff_options *o, const char *set, const char *reset,
466                       const char *line, int len)
467 {
468         emit_line_0(o, set, reset, line[0], line+1, len-1);
469 }
470
471 static int new_blank_line_at_eof(struct emit_callback *ecbdata, const char *line, int len)
472 {
473         if (!((ecbdata->ws_rule & WS_BLANK_AT_EOF) &&
474               ecbdata->blank_at_eof_in_preimage &&
475               ecbdata->blank_at_eof_in_postimage &&
476               ecbdata->blank_at_eof_in_preimage <= ecbdata->lno_in_preimage &&
477               ecbdata->blank_at_eof_in_postimage <= ecbdata->lno_in_postimage))
478                 return 0;
479         return ws_blank_line(line, len, ecbdata->ws_rule);
480 }
481
482 static void emit_add_line(const char *reset,
483                           struct emit_callback *ecbdata,
484                           const char *line, int len)
485 {
486         const char *ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
487         const char *set = diff_get_color(ecbdata->color_diff, DIFF_FILE_NEW);
488
489         if (!*ws)
490                 emit_line_0(ecbdata->opt, set, reset, '+', line, len);
491         else if (new_blank_line_at_eof(ecbdata, line, len))
492                 /* Blank line at EOF - paint '+' as well */
493                 emit_line_0(ecbdata->opt, ws, reset, '+', line, len);
494         else {
495                 /* Emit just the prefix, then the rest. */
496                 emit_line_0(ecbdata->opt, set, reset, '+', "", 0);
497                 ws_check_emit(line, len, ecbdata->ws_rule,
498                               ecbdata->opt->file, set, reset, ws);
499         }
500 }
501
502 static void emit_hunk_header(struct emit_callback *ecbdata,
503                              const char *line, int len)
504 {
505         const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
506         const char *frag = diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO);
507         const char *func = diff_get_color(ecbdata->color_diff, DIFF_FUNCINFO);
508         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
509         static const char atat[2] = { '@', '@' };
510         const char *cp, *ep;
511         struct strbuf msgbuf = STRBUF_INIT;
512         int org_len = len;
513         int i = 1;
514
515         /*
516          * As a hunk header must begin with "@@ -<old>, +<new> @@",
517          * it always is at least 10 bytes long.
518          */
519         if (len < 10 ||
520             memcmp(line, atat, 2) ||
521             !(ep = memmem(line + 2, len - 2, atat, 2))) {
522                 emit_line(ecbdata->opt, plain, reset, line, len);
523                 return;
524         }
525         ep += 2; /* skip over @@ */
526
527         /* The hunk header in fraginfo color */
528         strbuf_addstr(&msgbuf, frag);
529         strbuf_add(&msgbuf, line, ep - line);
530         strbuf_addstr(&msgbuf, reset);
531
532         /*
533          * trailing "\r\n"
534          */
535         for ( ; i < 3; i++)
536                 if (line[len - i] == '\r' || line[len - i] == '\n')
537                         len--;
538
539         /* blank before the func header */
540         for (cp = ep; ep - line < len; ep++)
541                 if (*ep != ' ' && *ep != '\t')
542                         break;
543         if (ep != cp) {
544                 strbuf_addstr(&msgbuf, plain);
545                 strbuf_add(&msgbuf, cp, ep - cp);
546                 strbuf_addstr(&msgbuf, reset);
547         }
548
549         if (ep < line + len) {
550                 strbuf_addstr(&msgbuf, func);
551                 strbuf_add(&msgbuf, ep, line + len - ep);
552                 strbuf_addstr(&msgbuf, reset);
553         }
554
555         strbuf_add(&msgbuf, line + len, org_len - len);
556         emit_line(ecbdata->opt, "", "", msgbuf.buf, msgbuf.len);
557         strbuf_release(&msgbuf);
558 }
559
560 static struct diff_tempfile *claim_diff_tempfile(void) {
561         int i;
562         for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
563                 if (!diff_temp[i].name)
564                         return diff_temp + i;
565         die("BUG: diff is failing to clean up its tempfiles");
566 }
567
568 static int remove_tempfile_installed;
569
570 static void remove_tempfile(void)
571 {
572         int i;
573         for (i = 0; i < ARRAY_SIZE(diff_temp); i++) {
574                 if (diff_temp[i].name == diff_temp[i].tmp_path)
575                         unlink_or_warn(diff_temp[i].name);
576                 diff_temp[i].name = NULL;
577         }
578 }
579
580 static void remove_tempfile_on_signal(int signo)
581 {
582         remove_tempfile();
583         sigchain_pop(signo);
584         raise(signo);
585 }
586
587 static void print_line_count(FILE *file, int count)
588 {
589         switch (count) {
590         case 0:
591                 fprintf(file, "0,0");
592                 break;
593         case 1:
594                 fprintf(file, "1");
595                 break;
596         default:
597                 fprintf(file, "1,%d", count);
598                 break;
599         }
600 }
601
602 static void emit_rewrite_lines(struct emit_callback *ecb,
603                                int prefix, const char *data, int size)
604 {
605         const char *endp = NULL;
606         static const char *nneof = " No newline at end of file\n";
607         const char *old = diff_get_color(ecb->color_diff, DIFF_FILE_OLD);
608         const char *reset = diff_get_color(ecb->color_diff, DIFF_RESET);
609
610         while (0 < size) {
611                 int len;
612
613                 endp = memchr(data, '\n', size);
614                 len = endp ? (endp - data + 1) : size;
615                 if (prefix != '+') {
616                         ecb->lno_in_preimage++;
617                         emit_line_0(ecb->opt, old, reset, '-',
618                                     data, len);
619                 } else {
620                         ecb->lno_in_postimage++;
621                         emit_add_line(reset, ecb, data, len);
622                 }
623                 size -= len;
624                 data += len;
625         }
626         if (!endp) {
627                 const char *plain = diff_get_color(ecb->color_diff,
628                                                    DIFF_PLAIN);
629                 putc('\n', ecb->opt->file);
630                 emit_line_0(ecb->opt, plain, reset, '\\',
631                             nneof, strlen(nneof));
632         }
633 }
634
635 static void emit_rewrite_diff(const char *name_a,
636                               const char *name_b,
637                               struct diff_filespec *one,
638                               struct diff_filespec *two,
639                               struct userdiff_driver *textconv_one,
640                               struct userdiff_driver *textconv_two,
641                               struct diff_options *o)
642 {
643         int lc_a, lc_b;
644         const char *name_a_tab, *name_b_tab;
645         const char *metainfo = diff_get_color(o->use_color, DIFF_METAINFO);
646         const char *fraginfo = diff_get_color(o->use_color, DIFF_FRAGINFO);
647         const char *reset = diff_get_color(o->use_color, DIFF_RESET);
648         static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
649         const char *a_prefix, *b_prefix;
650         char *data_one, *data_two;
651         size_t size_one, size_two;
652         struct emit_callback ecbdata;
653         const char *line_prefix = diff_line_prefix(o);
654
655         if (diff_mnemonic_prefix && DIFF_OPT_TST(o, REVERSE_DIFF)) {
656                 a_prefix = o->b_prefix;
657                 b_prefix = o->a_prefix;
658         } else {
659                 a_prefix = o->a_prefix;
660                 b_prefix = o->b_prefix;
661         }
662
663         name_a += (*name_a == '/');
664         name_b += (*name_b == '/');
665         name_a_tab = strchr(name_a, ' ') ? "\t" : "";
666         name_b_tab = strchr(name_b, ' ') ? "\t" : "";
667
668         strbuf_reset(&a_name);
669         strbuf_reset(&b_name);
670         quote_two_c_style(&a_name, a_prefix, name_a, 0);
671         quote_two_c_style(&b_name, b_prefix, name_b, 0);
672
673         size_one = fill_textconv(textconv_one, one, &data_one);
674         size_two = fill_textconv(textconv_two, two, &data_two);
675
676         memset(&ecbdata, 0, sizeof(ecbdata));
677         ecbdata.color_diff = want_color(o->use_color);
678         ecbdata.found_changesp = &o->found_changes;
679         ecbdata.ws_rule = whitespace_rule(name_b);
680         ecbdata.opt = o;
681         if (ecbdata.ws_rule & WS_BLANK_AT_EOF) {
682                 mmfile_t mf1, mf2;
683                 mf1.ptr = (char *)data_one;
684                 mf2.ptr = (char *)data_two;
685                 mf1.size = size_one;
686                 mf2.size = size_two;
687                 check_blank_at_eof(&mf1, &mf2, &ecbdata);
688         }
689         ecbdata.lno_in_preimage = 1;
690         ecbdata.lno_in_postimage = 1;
691
692         lc_a = count_lines(data_one, size_one);
693         lc_b = count_lines(data_two, size_two);
694         fprintf(o->file,
695                 "%s%s--- %s%s%s\n%s%s+++ %s%s%s\n%s%s@@ -",
696                 line_prefix, metainfo, a_name.buf, name_a_tab, reset,
697                 line_prefix, metainfo, b_name.buf, name_b_tab, reset,
698                 line_prefix, fraginfo);
699         if (!o->irreversible_delete)
700                 print_line_count(o->file, lc_a);
701         else
702                 fprintf(o->file, "?,?");
703         fprintf(o->file, " +");
704         print_line_count(o->file, lc_b);
705         fprintf(o->file, " @@%s\n", reset);
706         if (lc_a && !o->irreversible_delete)
707                 emit_rewrite_lines(&ecbdata, '-', data_one, size_one);
708         if (lc_b)
709                 emit_rewrite_lines(&ecbdata, '+', data_two, size_two);
710         if (textconv_one)
711                 free((char *)data_one);
712         if (textconv_two)
713                 free((char *)data_two);
714 }
715
716 struct diff_words_buffer {
717         mmfile_t text;
718         long alloc;
719         struct diff_words_orig {
720                 const char *begin, *end;
721         } *orig;
722         int orig_nr, orig_alloc;
723 };
724
725 static void diff_words_append(char *line, unsigned long len,
726                 struct diff_words_buffer *buffer)
727 {
728         ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
729         line++;
730         len--;
731         memcpy(buffer->text.ptr + buffer->text.size, line, len);
732         buffer->text.size += len;
733         buffer->text.ptr[buffer->text.size] = '\0';
734 }
735
736 struct diff_words_style_elem {
737         const char *prefix;
738         const char *suffix;
739         const char *color; /* NULL; filled in by the setup code if
740                             * color is enabled */
741 };
742
743 struct diff_words_style {
744         enum diff_words_type type;
745         struct diff_words_style_elem new, old, ctx;
746         const char *newline;
747 };
748
749 static struct diff_words_style diff_words_styles[] = {
750         { DIFF_WORDS_PORCELAIN, {"+", "\n"}, {"-", "\n"}, {" ", "\n"}, "~\n" },
751         { DIFF_WORDS_PLAIN, {"{+", "+}"}, {"[-", "-]"}, {"", ""}, "\n" },
752         { DIFF_WORDS_COLOR, {"", ""}, {"", ""}, {"", ""}, "\n" }
753 };
754
755 struct diff_words_data {
756         struct diff_words_buffer minus, plus;
757         const char *current_plus;
758         int last_minus;
759         struct diff_options *opt;
760         regex_t *word_regex;
761         enum diff_words_type type;
762         struct diff_words_style *style;
763 };
764
765 static int fn_out_diff_words_write_helper(FILE *fp,
766                                           struct diff_words_style_elem *st_el,
767                                           const char *newline,
768                                           size_t count, const char *buf,
769                                           const char *line_prefix)
770 {
771         int print = 0;
772
773         while (count) {
774                 char *p = memchr(buf, '\n', count);
775                 if (print)
776                         fputs(line_prefix, fp);
777                 if (p != buf) {
778                         if (st_el->color && fputs(st_el->color, fp) < 0)
779                                 return -1;
780                         if (fputs(st_el->prefix, fp) < 0 ||
781                             fwrite(buf, p ? p - buf : count, 1, fp) != 1 ||
782                             fputs(st_el->suffix, fp) < 0)
783                                 return -1;
784                         if (st_el->color && *st_el->color
785                             && fputs(GIT_COLOR_RESET, fp) < 0)
786                                 return -1;
787                 }
788                 if (!p)
789                         return 0;
790                 if (fputs(newline, fp) < 0)
791                         return -1;
792                 count -= p + 1 - buf;
793                 buf = p + 1;
794                 print = 1;
795         }
796         return 0;
797 }
798
799 /*
800  * '--color-words' algorithm can be described as:
801  *
802  *   1. collect a the minus/plus lines of a diff hunk, divided into
803  *      minus-lines and plus-lines;
804  *
805  *   2. break both minus-lines and plus-lines into words and
806  *      place them into two mmfile_t with one word for each line;
807  *
808  *   3. use xdiff to run diff on the two mmfile_t to get the words level diff;
809  *
810  * And for the common parts of the both file, we output the plus side text.
811  * diff_words->current_plus is used to trace the current position of the plus file
812  * which printed. diff_words->last_minus is used to trace the last minus word
813  * printed.
814  *
815  * For '--graph' to work with '--color-words', we need to output the graph prefix
816  * on each line of color words output. Generally, there are two conditions on
817  * which we should output the prefix.
818  *
819  *   1. diff_words->last_minus == 0 &&
820  *      diff_words->current_plus == diff_words->plus.text.ptr
821  *
822  *      that is: the plus text must start as a new line, and if there is no minus
823  *      word printed, a graph prefix must be printed.
824  *
825  *   2. diff_words->current_plus > diff_words->plus.text.ptr &&
826  *      *(diff_words->current_plus - 1) == '\n'
827  *
828  *      that is: a graph prefix must be printed following a '\n'
829  */
830 static int color_words_output_graph_prefix(struct diff_words_data *diff_words)
831 {
832         if ((diff_words->last_minus == 0 &&
833                 diff_words->current_plus == diff_words->plus.text.ptr) ||
834                 (diff_words->current_plus > diff_words->plus.text.ptr &&
835                 *(diff_words->current_plus - 1) == '\n')) {
836                 return 1;
837         } else {
838                 return 0;
839         }
840 }
841
842 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
843 {
844         struct diff_words_data *diff_words = priv;
845         struct diff_words_style *style = diff_words->style;
846         int minus_first, minus_len, plus_first, plus_len;
847         const char *minus_begin, *minus_end, *plus_begin, *plus_end;
848         struct diff_options *opt = diff_words->opt;
849         const char *line_prefix;
850
851         if (line[0] != '@' || parse_hunk_header(line, len,
852                         &minus_first, &minus_len, &plus_first, &plus_len))
853                 return;
854
855         assert(opt);
856         line_prefix = diff_line_prefix(opt);
857
858         /* POSIX requires that first be decremented by one if len == 0... */
859         if (minus_len) {
860                 minus_begin = diff_words->minus.orig[minus_first].begin;
861                 minus_end =
862                         diff_words->minus.orig[minus_first + minus_len - 1].end;
863         } else
864                 minus_begin = minus_end =
865                         diff_words->minus.orig[minus_first].end;
866
867         if (plus_len) {
868                 plus_begin = diff_words->plus.orig[plus_first].begin;
869                 plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
870         } else
871                 plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
872
873         if (color_words_output_graph_prefix(diff_words)) {
874                 fputs(line_prefix, diff_words->opt->file);
875         }
876         if (diff_words->current_plus != plus_begin) {
877                 fn_out_diff_words_write_helper(diff_words->opt->file,
878                                 &style->ctx, style->newline,
879                                 plus_begin - diff_words->current_plus,
880                                 diff_words->current_plus, line_prefix);
881                 if (*(plus_begin - 1) == '\n')
882                         fputs(line_prefix, diff_words->opt->file);
883         }
884         if (minus_begin != minus_end) {
885                 fn_out_diff_words_write_helper(diff_words->opt->file,
886                                 &style->old, style->newline,
887                                 minus_end - minus_begin, minus_begin,
888                                 line_prefix);
889         }
890         if (plus_begin != plus_end) {
891                 fn_out_diff_words_write_helper(diff_words->opt->file,
892                                 &style->new, style->newline,
893                                 plus_end - plus_begin, plus_begin,
894                                 line_prefix);
895         }
896
897         diff_words->current_plus = plus_end;
898         diff_words->last_minus = minus_first;
899 }
900
901 /* This function starts looking at *begin, and returns 0 iff a word was found. */
902 static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
903                 int *begin, int *end)
904 {
905         if (word_regex && *begin < buffer->size) {
906                 regmatch_t match[1];
907                 if (!regexec(word_regex, buffer->ptr + *begin, 1, match, 0)) {
908                         char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
909                                         '\n', match[0].rm_eo - match[0].rm_so);
910                         *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
911                         *begin += match[0].rm_so;
912                         return *begin >= *end;
913                 }
914                 return -1;
915         }
916
917         /* find the next word */
918         while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
919                 (*begin)++;
920         if (*begin >= buffer->size)
921                 return -1;
922
923         /* find the end of the word */
924         *end = *begin + 1;
925         while (*end < buffer->size && !isspace(buffer->ptr[*end]))
926                 (*end)++;
927
928         return 0;
929 }
930
931 /*
932  * This function splits the words in buffer->text, stores the list with
933  * newline separator into out, and saves the offsets of the original words
934  * in buffer->orig.
935  */
936 static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
937                 regex_t *word_regex)
938 {
939         int i, j;
940         long alloc = 0;
941
942         out->size = 0;
943         out->ptr = NULL;
944
945         /* fake an empty "0th" word */
946         ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
947         buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
948         buffer->orig_nr = 1;
949
950         for (i = 0; i < buffer->text.size; i++) {
951                 if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
952                         return;
953
954                 /* store original boundaries */
955                 ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
956                                 buffer->orig_alloc);
957                 buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
958                 buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
959                 buffer->orig_nr++;
960
961                 /* store one word */
962                 ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
963                 memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
964                 out->ptr[out->size + j - i] = '\n';
965                 out->size += j - i + 1;
966
967                 i = j - 1;
968         }
969 }
970
971 /* this executes the word diff on the accumulated buffers */
972 static void diff_words_show(struct diff_words_data *diff_words)
973 {
974         xpparam_t xpp;
975         xdemitconf_t xecfg;
976         mmfile_t minus, plus;
977         struct diff_words_style *style = diff_words->style;
978
979         struct diff_options *opt = diff_words->opt;
980         const char *line_prefix;
981
982         assert(opt);
983         line_prefix = diff_line_prefix(opt);
984
985         /* special case: only removal */
986         if (!diff_words->plus.text.size) {
987                 fputs(line_prefix, diff_words->opt->file);
988                 fn_out_diff_words_write_helper(diff_words->opt->file,
989                         &style->old, style->newline,
990                         diff_words->minus.text.size,
991                         diff_words->minus.text.ptr, line_prefix);
992                 diff_words->minus.text.size = 0;
993                 return;
994         }
995
996         diff_words->current_plus = diff_words->plus.text.ptr;
997         diff_words->last_minus = 0;
998
999         memset(&xpp, 0, sizeof(xpp));
1000         memset(&xecfg, 0, sizeof(xecfg));
1001         diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
1002         diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
1003         xpp.flags = 0;
1004         /* as only the hunk header will be parsed, we need a 0-context */
1005         xecfg.ctxlen = 0;
1006         xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, diff_words,
1007                       &xpp, &xecfg);
1008         free(minus.ptr);
1009         free(plus.ptr);
1010         if (diff_words->current_plus != diff_words->plus.text.ptr +
1011                         diff_words->plus.text.size) {
1012                 if (color_words_output_graph_prefix(diff_words))
1013                         fputs(line_prefix, diff_words->opt->file);
1014                 fn_out_diff_words_write_helper(diff_words->opt->file,
1015                         &style->ctx, style->newline,
1016                         diff_words->plus.text.ptr + diff_words->plus.text.size
1017                         - diff_words->current_plus, diff_words->current_plus,
1018                         line_prefix);
1019         }
1020         diff_words->minus.text.size = diff_words->plus.text.size = 0;
1021 }
1022
1023 /* In "color-words" mode, show word-diff of words accumulated in the buffer */
1024 static void diff_words_flush(struct emit_callback *ecbdata)
1025 {
1026         if (ecbdata->diff_words->minus.text.size ||
1027             ecbdata->diff_words->plus.text.size)
1028                 diff_words_show(ecbdata->diff_words);
1029 }
1030
1031 static void diff_filespec_load_driver(struct diff_filespec *one)
1032 {
1033         /* Use already-loaded driver */
1034         if (one->driver)
1035                 return;
1036
1037         if (S_ISREG(one->mode))
1038                 one->driver = userdiff_find_by_path(one->path);
1039
1040         /* Fallback to default settings */
1041         if (!one->driver)
1042                 one->driver = userdiff_find_by_name("default");
1043 }
1044
1045 static const char *userdiff_word_regex(struct diff_filespec *one)
1046 {
1047         diff_filespec_load_driver(one);
1048         return one->driver->word_regex;
1049 }
1050
1051 static void init_diff_words_data(struct emit_callback *ecbdata,
1052                                  struct diff_options *orig_opts,
1053                                  struct diff_filespec *one,
1054                                  struct diff_filespec *two)
1055 {
1056         int i;
1057         struct diff_options *o = xmalloc(sizeof(struct diff_options));
1058         memcpy(o, orig_opts, sizeof(struct diff_options));
1059
1060         ecbdata->diff_words =
1061                 xcalloc(1, sizeof(struct diff_words_data));
1062         ecbdata->diff_words->type = o->word_diff;
1063         ecbdata->diff_words->opt = o;
1064         if (!o->word_regex)
1065                 o->word_regex = userdiff_word_regex(one);
1066         if (!o->word_regex)
1067                 o->word_regex = userdiff_word_regex(two);
1068         if (!o->word_regex)
1069                 o->word_regex = diff_word_regex_cfg;
1070         if (o->word_regex) {
1071                 ecbdata->diff_words->word_regex = (regex_t *)
1072                         xmalloc(sizeof(regex_t));
1073                 if (regcomp(ecbdata->diff_words->word_regex,
1074                             o->word_regex,
1075                             REG_EXTENDED | REG_NEWLINE))
1076                         die ("Invalid regular expression: %s",
1077                              o->word_regex);
1078         }
1079         for (i = 0; i < ARRAY_SIZE(diff_words_styles); i++) {
1080                 if (o->word_diff == diff_words_styles[i].type) {
1081                         ecbdata->diff_words->style =
1082                                 &diff_words_styles[i];
1083                         break;
1084                 }
1085         }
1086         if (want_color(o->use_color)) {
1087                 struct diff_words_style *st = ecbdata->diff_words->style;
1088                 st->old.color = diff_get_color_opt(o, DIFF_FILE_OLD);
1089                 st->new.color = diff_get_color_opt(o, DIFF_FILE_NEW);
1090                 st->ctx.color = diff_get_color_opt(o, DIFF_PLAIN);
1091         }
1092 }
1093
1094 static void free_diff_words_data(struct emit_callback *ecbdata)
1095 {
1096         if (ecbdata->diff_words) {
1097                 diff_words_flush(ecbdata);
1098                 free (ecbdata->diff_words->opt);
1099                 free (ecbdata->diff_words->minus.text.ptr);
1100                 free (ecbdata->diff_words->minus.orig);
1101                 free (ecbdata->diff_words->plus.text.ptr);
1102                 free (ecbdata->diff_words->plus.orig);
1103                 if (ecbdata->diff_words->word_regex) {
1104                         regfree(ecbdata->diff_words->word_regex);
1105                         free(ecbdata->diff_words->word_regex);
1106                 }
1107                 free(ecbdata->diff_words);
1108                 ecbdata->diff_words = NULL;
1109         }
1110 }
1111
1112 const char *diff_get_color(int diff_use_color, enum color_diff ix)
1113 {
1114         if (want_color(diff_use_color))
1115                 return diff_colors[ix];
1116         return "";
1117 }
1118
1119 const char *diff_line_prefix(struct diff_options *opt)
1120 {
1121         struct strbuf *msgbuf;
1122         if (!opt->output_prefix)
1123                 return "";
1124
1125         msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
1126         return msgbuf->buf;
1127 }
1128
1129 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
1130 {
1131         const char *cp;
1132         unsigned long allot;
1133         size_t l = len;
1134
1135         if (ecb->truncate)
1136                 return ecb->truncate(line, len);
1137         cp = line;
1138         allot = l;
1139         while (0 < l) {
1140                 (void) utf8_width(&cp, &l);
1141                 if (!cp)
1142                         break; /* truncated in the middle? */
1143         }
1144         return allot - l;
1145 }
1146
1147 static void find_lno(const char *line, struct emit_callback *ecbdata)
1148 {
1149         const char *p;
1150         ecbdata->lno_in_preimage = 0;
1151         ecbdata->lno_in_postimage = 0;
1152         p = strchr(line, '-');
1153         if (!p)
1154                 return; /* cannot happen */
1155         ecbdata->lno_in_preimage = strtol(p + 1, NULL, 10);
1156         p = strchr(p, '+');
1157         if (!p)
1158                 return; /* cannot happen */
1159         ecbdata->lno_in_postimage = strtol(p + 1, NULL, 10);
1160 }
1161
1162 static void fn_out_consume(void *priv, char *line, unsigned long len)
1163 {
1164         struct emit_callback *ecbdata = priv;
1165         const char *meta = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
1166         const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
1167         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
1168         struct diff_options *o = ecbdata->opt;
1169         const char *line_prefix = diff_line_prefix(o);
1170
1171         if (ecbdata->header) {
1172                 fprintf(ecbdata->opt->file, "%s", ecbdata->header->buf);
1173                 strbuf_reset(ecbdata->header);
1174                 ecbdata->header = NULL;
1175         }
1176         *(ecbdata->found_changesp) = 1;
1177
1178         if (ecbdata->label_path[0]) {
1179                 const char *name_a_tab, *name_b_tab;
1180
1181                 name_a_tab = strchr(ecbdata->label_path[0], ' ') ? "\t" : "";
1182                 name_b_tab = strchr(ecbdata->label_path[1], ' ') ? "\t" : "";
1183
1184                 fprintf(ecbdata->opt->file, "%s%s--- %s%s%s\n",
1185                         line_prefix, meta, ecbdata->label_path[0], reset, name_a_tab);
1186                 fprintf(ecbdata->opt->file, "%s%s+++ %s%s%s\n",
1187                         line_prefix, meta, ecbdata->label_path[1], reset, name_b_tab);
1188                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
1189         }
1190
1191         if (diff_suppress_blank_empty
1192             && len == 2 && line[0] == ' ' && line[1] == '\n') {
1193                 line[0] = '\n';
1194                 len = 1;
1195         }
1196
1197         if (line[0] == '@') {
1198                 if (ecbdata->diff_words)
1199                         diff_words_flush(ecbdata);
1200                 len = sane_truncate_line(ecbdata, line, len);
1201                 find_lno(line, ecbdata);
1202                 emit_hunk_header(ecbdata, line, len);
1203                 if (line[len-1] != '\n')
1204                         putc('\n', ecbdata->opt->file);
1205                 return;
1206         }
1207
1208         if (len < 1) {
1209                 emit_line(ecbdata->opt, reset, reset, line, len);
1210                 if (ecbdata->diff_words
1211                     && ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN)
1212                         fputs("~\n", ecbdata->opt->file);
1213                 return;
1214         }
1215
1216         if (ecbdata->diff_words) {
1217                 if (line[0] == '-') {
1218                         diff_words_append(line, len,
1219                                           &ecbdata->diff_words->minus);
1220                         return;
1221                 } else if (line[0] == '+') {
1222                         diff_words_append(line, len,
1223                                           &ecbdata->diff_words->plus);
1224                         return;
1225                 } else if (starts_with(line, "\\ ")) {
1226                         /*
1227                          * Eat the "no newline at eof" marker as if we
1228                          * saw a "+" or "-" line with nothing on it,
1229                          * and return without diff_words_flush() to
1230                          * defer processing. If this is the end of
1231                          * preimage, more "+" lines may come after it.
1232                          */
1233                         return;
1234                 }
1235                 diff_words_flush(ecbdata);
1236                 if (ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN) {
1237                         emit_line(ecbdata->opt, plain, reset, line, len);
1238                         fputs("~\n", ecbdata->opt->file);
1239                 } else {
1240                         /*
1241                          * Skip the prefix character, if any.  With
1242                          * diff_suppress_blank_empty, there may be
1243                          * none.
1244                          */
1245                         if (line[0] != '\n') {
1246                               line++;
1247                               len--;
1248                         }
1249                         emit_line(ecbdata->opt, plain, reset, line, len);
1250                 }
1251                 return;
1252         }
1253
1254         if (line[0] != '+') {
1255                 const char *color =
1256                         diff_get_color(ecbdata->color_diff,
1257                                        line[0] == '-' ? DIFF_FILE_OLD : DIFF_PLAIN);
1258                 ecbdata->lno_in_preimage++;
1259                 if (line[0] == ' ')
1260                         ecbdata->lno_in_postimage++;
1261                 emit_line(ecbdata->opt, color, reset, line, len);
1262         } else {
1263                 ecbdata->lno_in_postimage++;
1264                 emit_add_line(reset, ecbdata, line + 1, len - 1);
1265         }
1266 }
1267
1268 static char *pprint_rename(const char *a, const char *b)
1269 {
1270         const char *old = a;
1271         const char *new = b;
1272         struct strbuf name = STRBUF_INIT;
1273         int pfx_length, sfx_length;
1274         int pfx_adjust_for_slash;
1275         int len_a = strlen(a);
1276         int len_b = strlen(b);
1277         int a_midlen, b_midlen;
1278         int qlen_a = quote_c_style(a, NULL, NULL, 0);
1279         int qlen_b = quote_c_style(b, NULL, NULL, 0);
1280
1281         if (qlen_a || qlen_b) {
1282                 quote_c_style(a, &name, NULL, 0);
1283                 strbuf_addstr(&name, " => ");
1284                 quote_c_style(b, &name, NULL, 0);
1285                 return strbuf_detach(&name, NULL);
1286         }
1287
1288         /* Find common prefix */
1289         pfx_length = 0;
1290         while (*old && *new && *old == *new) {
1291                 if (*old == '/')
1292                         pfx_length = old - a + 1;
1293                 old++;
1294                 new++;
1295         }
1296
1297         /* Find common suffix */
1298         old = a + len_a;
1299         new = b + len_b;
1300         sfx_length = 0;
1301         /*
1302          * If there is a common prefix, it must end in a slash.  In
1303          * that case we let this loop run 1 into the prefix to see the
1304          * same slash.
1305          *
1306          * If there is no common prefix, we cannot do this as it would
1307          * underrun the input strings.
1308          */
1309         pfx_adjust_for_slash = (pfx_length ? 1 : 0);
1310         while (a + pfx_length - pfx_adjust_for_slash <= old &&
1311                b + pfx_length - pfx_adjust_for_slash <= new &&
1312                *old == *new) {
1313                 if (*old == '/')
1314                         sfx_length = len_a - (old - a);
1315                 old--;
1316                 new--;
1317         }
1318
1319         /*
1320          * pfx{mid-a => mid-b}sfx
1321          * {pfx-a => pfx-b}sfx
1322          * pfx{sfx-a => sfx-b}
1323          * name-a => name-b
1324          */
1325         a_midlen = len_a - pfx_length - sfx_length;
1326         b_midlen = len_b - pfx_length - sfx_length;
1327         if (a_midlen < 0)
1328                 a_midlen = 0;
1329         if (b_midlen < 0)
1330                 b_midlen = 0;
1331
1332         strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
1333         if (pfx_length + sfx_length) {
1334                 strbuf_add(&name, a, pfx_length);
1335                 strbuf_addch(&name, '{');
1336         }
1337         strbuf_add(&name, a + pfx_length, a_midlen);
1338         strbuf_addstr(&name, " => ");
1339         strbuf_add(&name, b + pfx_length, b_midlen);
1340         if (pfx_length + sfx_length) {
1341                 strbuf_addch(&name, '}');
1342                 strbuf_add(&name, a + len_a - sfx_length, sfx_length);
1343         }
1344         return strbuf_detach(&name, NULL);
1345 }
1346
1347 struct diffstat_t {
1348         int nr;
1349         int alloc;
1350         struct diffstat_file {
1351                 char *from_name;
1352                 char *name;
1353                 char *print_name;
1354                 unsigned is_unmerged:1;
1355                 unsigned is_binary:1;
1356                 unsigned is_renamed:1;
1357                 unsigned is_interesting:1;
1358                 uintmax_t added, deleted;
1359         } **files;
1360 };
1361
1362 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
1363                                           const char *name_a,
1364                                           const char *name_b)
1365 {
1366         struct diffstat_file *x;
1367         x = xcalloc(1, sizeof(*x));
1368         ALLOC_GROW(diffstat->files, diffstat->nr + 1, diffstat->alloc);
1369         diffstat->files[diffstat->nr++] = x;
1370         if (name_b) {
1371                 x->from_name = xstrdup(name_a);
1372                 x->name = xstrdup(name_b);
1373                 x->is_renamed = 1;
1374         }
1375         else {
1376                 x->from_name = NULL;
1377                 x->name = xstrdup(name_a);
1378         }
1379         return x;
1380 }
1381
1382 static void diffstat_consume(void *priv, char *line, unsigned long len)
1383 {
1384         struct diffstat_t *diffstat = priv;
1385         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
1386
1387         if (line[0] == '+')
1388                 x->added++;
1389         else if (line[0] == '-')
1390                 x->deleted++;
1391 }
1392
1393 const char mime_boundary_leader[] = "------------";
1394
1395 static int scale_linear(int it, int width, int max_change)
1396 {
1397         if (!it)
1398                 return 0;
1399         /*
1400          * make sure that at least one '-' or '+' is printed if
1401          * there is any change to this path. The easiest way is to
1402          * scale linearly as if the alloted width is one column shorter
1403          * than it is, and then add 1 to the result.
1404          */
1405         return 1 + (it * (width - 1) / max_change);
1406 }
1407
1408 static void show_name(FILE *file,
1409                       const char *prefix, const char *name, int len)
1410 {
1411         fprintf(file, " %s%-*s |", prefix, len, name);
1412 }
1413
1414 static void show_graph(FILE *file, char ch, int cnt, const char *set, const char *reset)
1415 {
1416         if (cnt <= 0)
1417                 return;
1418         fprintf(file, "%s", set);
1419         while (cnt--)
1420                 putc(ch, file);
1421         fprintf(file, "%s", reset);
1422 }
1423
1424 static void fill_print_name(struct diffstat_file *file)
1425 {
1426         char *pname;
1427
1428         if (file->print_name)
1429                 return;
1430
1431         if (!file->is_renamed) {
1432                 struct strbuf buf = STRBUF_INIT;
1433                 if (quote_c_style(file->name, &buf, NULL, 0)) {
1434                         pname = strbuf_detach(&buf, NULL);
1435                 } else {
1436                         pname = file->name;
1437                         strbuf_release(&buf);
1438                 }
1439         } else {
1440                 pname = pprint_rename(file->from_name, file->name);
1441         }
1442         file->print_name = pname;
1443 }
1444
1445 int print_stat_summary(FILE *fp, int files, int insertions, int deletions)
1446 {
1447         struct strbuf sb = STRBUF_INIT;
1448         int ret;
1449
1450         if (!files) {
1451                 assert(insertions == 0 && deletions == 0);
1452                 return fprintf(fp, "%s\n", " 0 files changed");
1453         }
1454
1455         strbuf_addf(&sb,
1456                     (files == 1) ? " %d file changed" : " %d files changed",
1457                     files);
1458
1459         /*
1460          * For binary diff, the caller may want to print "x files
1461          * changed" with insertions == 0 && deletions == 0.
1462          *
1463          * Not omitting "0 insertions(+), 0 deletions(-)" in this case
1464          * is probably less confusing (i.e skip over "2 files changed
1465          * but nothing about added/removed lines? Is this a bug in Git?").
1466          */
1467         if (insertions || deletions == 0) {
1468                 strbuf_addf(&sb,
1469                             (insertions == 1) ? ", %d insertion(+)" : ", %d insertions(+)",
1470                             insertions);
1471         }
1472
1473         if (deletions || insertions == 0) {
1474                 strbuf_addf(&sb,
1475                             (deletions == 1) ? ", %d deletion(-)" : ", %d deletions(-)",
1476                             deletions);
1477         }
1478         strbuf_addch(&sb, '\n');
1479         ret = fputs(sb.buf, fp);
1480         strbuf_release(&sb);
1481         return ret;
1482 }
1483
1484 static void show_stats(struct diffstat_t *data, struct diff_options *options)
1485 {
1486         int i, len, add, del, adds = 0, dels = 0;
1487         uintmax_t max_change = 0, max_len = 0;
1488         int total_files = data->nr, count;
1489         int width, name_width, graph_width, number_width = 0, bin_width = 0;
1490         const char *reset, *add_c, *del_c;
1491         const char *line_prefix = "";
1492         int extra_shown = 0;
1493
1494         if (data->nr == 0)
1495                 return;
1496
1497         line_prefix = diff_line_prefix(options);
1498         count = options->stat_count ? options->stat_count : data->nr;
1499
1500         reset = diff_get_color_opt(options, DIFF_RESET);
1501         add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
1502         del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
1503
1504         /*
1505          * Find the longest filename and max number of changes
1506          */
1507         for (i = 0; (i < count) && (i < data->nr); i++) {
1508                 struct diffstat_file *file = data->files[i];
1509                 uintmax_t change = file->added + file->deleted;
1510
1511                 if (!file->is_interesting && (change == 0)) {
1512                         count++; /* not shown == room for one more */
1513                         continue;
1514                 }
1515                 fill_print_name(file);
1516                 len = strlen(file->print_name);
1517                 if (max_len < len)
1518                         max_len = len;
1519
1520                 if (file->is_unmerged) {
1521                         /* "Unmerged" is 8 characters */
1522                         bin_width = bin_width < 8 ? 8 : bin_width;
1523                         continue;
1524                 }
1525                 if (file->is_binary) {
1526                         /* "Bin XXX -> YYY bytes" */
1527                         int w = 14 + decimal_width(file->added)
1528                                 + decimal_width(file->deleted);
1529                         bin_width = bin_width < w ? w : bin_width;
1530                         /* Display change counts aligned with "Bin" */
1531                         number_width = 3;
1532                         continue;
1533                 }
1534
1535                 if (max_change < change)
1536                         max_change = change;
1537         }
1538         count = i; /* where we can stop scanning in data->files[] */
1539
1540         /*
1541          * We have width = stat_width or term_columns() columns total.
1542          * We want a maximum of min(max_len, stat_name_width) for the name part.
1543          * We want a maximum of min(max_change, stat_graph_width) for the +- part.
1544          * We also need 1 for " " and 4 + decimal_width(max_change)
1545          * for " | NNNN " and one the empty column at the end, altogether
1546          * 6 + decimal_width(max_change).
1547          *
1548          * If there's not enough space, we will use the smaller of
1549          * stat_name_width (if set) and 5/8*width for the filename,
1550          * and the rest for constant elements + graph part, but no more
1551          * than stat_graph_width for the graph part.
1552          * (5/8 gives 50 for filename and 30 for the constant parts + graph
1553          * for the standard terminal size).
1554          *
1555          * In other words: stat_width limits the maximum width, and
1556          * stat_name_width fixes the maximum width of the filename,
1557          * and is also used to divide available columns if there
1558          * aren't enough.
1559          *
1560          * Binary files are displayed with "Bin XXX -> YYY bytes"
1561          * instead of the change count and graph. This part is treated
1562          * similarly to the graph part, except that it is not
1563          * "scaled". If total width is too small to accommodate the
1564          * guaranteed minimum width of the filename part and the
1565          * separators and this message, this message will "overflow"
1566          * making the line longer than the maximum width.
1567          */
1568
1569         if (options->stat_width == -1)
1570                 width = term_columns() - options->output_prefix_length;
1571         else
1572                 width = options->stat_width ? options->stat_width : 80;
1573         number_width = decimal_width(max_change) > number_width ?
1574                 decimal_width(max_change) : number_width;
1575
1576         if (options->stat_graph_width == -1)
1577                 options->stat_graph_width = diff_stat_graph_width;
1578
1579         /*
1580          * Guarantee 3/8*16==6 for the graph part
1581          * and 5/8*16==10 for the filename part
1582          */
1583         if (width < 16 + 6 + number_width)
1584                 width = 16 + 6 + number_width;
1585
1586         /*
1587          * First assign sizes that are wanted, ignoring available width.
1588          * strlen("Bin XXX -> YYY bytes") == bin_width, and the part
1589          * starting from "XXX" should fit in graph_width.
1590          */
1591         graph_width = max_change + 4 > bin_width ? max_change : bin_width - 4;
1592         if (options->stat_graph_width &&
1593             options->stat_graph_width < graph_width)
1594                 graph_width = options->stat_graph_width;
1595
1596         name_width = (options->stat_name_width > 0 &&
1597                       options->stat_name_width < max_len) ?
1598                 options->stat_name_width : max_len;
1599
1600         /*
1601          * Adjust adjustable widths not to exceed maximum width
1602          */
1603         if (name_width + number_width + 6 + graph_width > width) {
1604                 if (graph_width > width * 3/8 - number_width - 6) {
1605                         graph_width = width * 3/8 - number_width - 6;
1606                         if (graph_width < 6)
1607                                 graph_width = 6;
1608                 }
1609
1610                 if (options->stat_graph_width &&
1611                     graph_width > options->stat_graph_width)
1612                         graph_width = options->stat_graph_width;
1613                 if (name_width > width - number_width - 6 - graph_width)
1614                         name_width = width - number_width - 6 - graph_width;
1615                 else
1616                         graph_width = width - number_width - 6 - name_width;
1617         }
1618
1619         /*
1620          * From here name_width is the width of the name area,
1621          * and graph_width is the width of the graph area.
1622          * max_change is used to scale graph properly.
1623          */
1624         for (i = 0; i < count; i++) {
1625                 const char *prefix = "";
1626                 struct diffstat_file *file = data->files[i];
1627                 char *name = file->print_name;
1628                 uintmax_t added = file->added;
1629                 uintmax_t deleted = file->deleted;
1630                 int name_len;
1631
1632                 if (!file->is_interesting && (added + deleted == 0))
1633                         continue;
1634
1635                 /*
1636                  * "scale" the filename
1637                  */
1638                 len = name_width;
1639                 name_len = strlen(name);
1640                 if (name_width < name_len) {
1641                         char *slash;
1642                         prefix = "...";
1643                         len -= 3;
1644                         name += name_len - len;
1645                         slash = strchr(name, '/');
1646                         if (slash)
1647                                 name = slash;
1648                 }
1649
1650                 if (file->is_binary) {
1651                         fprintf(options->file, "%s", line_prefix);
1652                         show_name(options->file, prefix, name, len);
1653                         fprintf(options->file, " %*s", number_width, "Bin");
1654                         if (!added && !deleted) {
1655                                 putc('\n', options->file);
1656                                 continue;
1657                         }
1658                         fprintf(options->file, " %s%"PRIuMAX"%s",
1659                                 del_c, deleted, reset);
1660                         fprintf(options->file, " -> ");
1661                         fprintf(options->file, "%s%"PRIuMAX"%s",
1662                                 add_c, added, reset);
1663                         fprintf(options->file, " bytes");
1664                         fprintf(options->file, "\n");
1665                         continue;
1666                 }
1667                 else if (file->is_unmerged) {
1668                         fprintf(options->file, "%s", line_prefix);
1669                         show_name(options->file, prefix, name, len);
1670                         fprintf(options->file, " Unmerged\n");
1671                         continue;
1672                 }
1673
1674                 /*
1675                  * scale the add/delete
1676                  */
1677                 add = added;
1678                 del = deleted;
1679
1680                 if (graph_width <= max_change) {
1681                         int total = scale_linear(add + del, graph_width, max_change);
1682                         if (total < 2 && add && del)
1683                                 /* width >= 2 due to the sanity check */
1684                                 total = 2;
1685                         if (add < del) {
1686                                 add = scale_linear(add, graph_width, max_change);
1687                                 del = total - add;
1688                         } else {
1689                                 del = scale_linear(del, graph_width, max_change);
1690                                 add = total - del;
1691                         }
1692                 }
1693                 fprintf(options->file, "%s", line_prefix);
1694                 show_name(options->file, prefix, name, len);
1695                 fprintf(options->file, " %*"PRIuMAX"%s",
1696                         number_width, added + deleted,
1697                         added + deleted ? " " : "");
1698                 show_graph(options->file, '+', add, add_c, reset);
1699                 show_graph(options->file, '-', del, del_c, reset);
1700                 fprintf(options->file, "\n");
1701         }
1702
1703         for (i = 0; i < data->nr; i++) {
1704                 struct diffstat_file *file = data->files[i];
1705                 uintmax_t added = file->added;
1706                 uintmax_t deleted = file->deleted;
1707
1708                 if (file->is_unmerged ||
1709                     (!file->is_interesting && (added + deleted == 0))) {
1710                         total_files--;
1711                         continue;
1712                 }
1713
1714                 if (!file->is_binary) {
1715                         adds += added;
1716                         dels += deleted;
1717                 }
1718                 if (i < count)
1719                         continue;
1720                 if (!extra_shown)
1721                         fprintf(options->file, "%s ...\n", line_prefix);
1722                 extra_shown = 1;
1723         }
1724         fprintf(options->file, "%s", line_prefix);
1725         print_stat_summary(options->file, total_files, adds, dels);
1726 }
1727
1728 static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
1729 {
1730         int i, adds = 0, dels = 0, total_files = data->nr;
1731
1732         if (data->nr == 0)
1733                 return;
1734
1735         for (i = 0; i < data->nr; i++) {
1736                 int added = data->files[i]->added;
1737                 int deleted= data->files[i]->deleted;
1738
1739                 if (data->files[i]->is_unmerged ||
1740                     (!data->files[i]->is_interesting && (added + deleted == 0))) {
1741                         total_files--;
1742                 } else if (!data->files[i]->is_binary) { /* don't count bytes */
1743                         adds += added;
1744                         dels += deleted;
1745                 }
1746         }
1747         fprintf(options->file, "%s", diff_line_prefix(options));
1748         print_stat_summary(options->file, total_files, adds, dels);
1749 }
1750
1751 static void show_numstat(struct diffstat_t *data, struct diff_options *options)
1752 {
1753         int i;
1754
1755         if (data->nr == 0)
1756                 return;
1757
1758         for (i = 0; i < data->nr; i++) {
1759                 struct diffstat_file *file = data->files[i];
1760
1761                 fprintf(options->file, "%s", diff_line_prefix(options));
1762
1763                 if (file->is_binary)
1764                         fprintf(options->file, "-\t-\t");
1765                 else
1766                         fprintf(options->file,
1767                                 "%"PRIuMAX"\t%"PRIuMAX"\t",
1768                                 file->added, file->deleted);
1769                 if (options->line_termination) {
1770                         fill_print_name(file);
1771                         if (!file->is_renamed)
1772                                 write_name_quoted(file->name, options->file,
1773                                                   options->line_termination);
1774                         else {
1775                                 fputs(file->print_name, options->file);
1776                                 putc(options->line_termination, options->file);
1777                         }
1778                 } else {
1779                         if (file->is_renamed) {
1780                                 putc('\0', options->file);
1781                                 write_name_quoted(file->from_name, options->file, '\0');
1782                         }
1783                         write_name_quoted(file->name, options->file, '\0');
1784                 }
1785         }
1786 }
1787
1788 struct dirstat_file {
1789         const char *name;
1790         unsigned long changed;
1791 };
1792
1793 struct dirstat_dir {
1794         struct dirstat_file *files;
1795         int alloc, nr, permille, cumulative;
1796 };
1797
1798 static long gather_dirstat(struct diff_options *opt, struct dirstat_dir *dir,
1799                 unsigned long changed, const char *base, int baselen)
1800 {
1801         unsigned long this_dir = 0;
1802         unsigned int sources = 0;
1803         const char *line_prefix = diff_line_prefix(opt);
1804
1805         while (dir->nr) {
1806                 struct dirstat_file *f = dir->files;
1807                 int namelen = strlen(f->name);
1808                 unsigned long this;
1809                 char *slash;
1810
1811                 if (namelen < baselen)
1812                         break;
1813                 if (memcmp(f->name, base, baselen))
1814                         break;
1815                 slash = strchr(f->name + baselen, '/');
1816                 if (slash) {
1817                         int newbaselen = slash + 1 - f->name;
1818                         this = gather_dirstat(opt, dir, changed, f->name, newbaselen);
1819                         sources++;
1820                 } else {
1821                         this = f->changed;
1822                         dir->files++;
1823                         dir->nr--;
1824                         sources += 2;
1825                 }
1826                 this_dir += this;
1827         }
1828
1829         /*
1830          * We don't report dirstat's for
1831          *  - the top level
1832          *  - or cases where everything came from a single directory
1833          *    under this directory (sources == 1).
1834          */
1835         if (baselen && sources != 1) {
1836                 if (this_dir) {
1837                         int permille = this_dir * 1000 / changed;
1838                         if (permille >= dir->permille) {
1839                                 fprintf(opt->file, "%s%4d.%01d%% %.*s\n", line_prefix,
1840                                         permille / 10, permille % 10, baselen, base);
1841                                 if (!dir->cumulative)
1842                                         return 0;
1843                         }
1844                 }
1845         }
1846         return this_dir;
1847 }
1848
1849 static int dirstat_compare(const void *_a, const void *_b)
1850 {
1851         const struct dirstat_file *a = _a;
1852         const struct dirstat_file *b = _b;
1853         return strcmp(a->name, b->name);
1854 }
1855
1856 static void show_dirstat(struct diff_options *options)
1857 {
1858         int i;
1859         unsigned long changed;
1860         struct dirstat_dir dir;
1861         struct diff_queue_struct *q = &diff_queued_diff;
1862
1863         dir.files = NULL;
1864         dir.alloc = 0;
1865         dir.nr = 0;
1866         dir.permille = options->dirstat_permille;
1867         dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
1868
1869         changed = 0;
1870         for (i = 0; i < q->nr; i++) {
1871                 struct diff_filepair *p = q->queue[i];
1872                 const char *name;
1873                 unsigned long copied, added, damage;
1874                 int content_changed;
1875
1876                 name = p->two->path ? p->two->path : p->one->path;
1877
1878                 if (p->one->sha1_valid && p->two->sha1_valid)
1879                         content_changed = hashcmp(p->one->sha1, p->two->sha1);
1880                 else
1881                         content_changed = 1;
1882
1883                 if (!content_changed) {
1884                         /*
1885                          * The SHA1 has not changed, so pre-/post-content is
1886                          * identical. We can therefore skip looking at the
1887                          * file contents altogether.
1888                          */
1889                         damage = 0;
1890                         goto found_damage;
1891                 }
1892
1893                 if (DIFF_OPT_TST(options, DIRSTAT_BY_FILE)) {
1894                         /*
1895                          * In --dirstat-by-file mode, we don't really need to
1896                          * look at the actual file contents at all.
1897                          * The fact that the SHA1 changed is enough for us to
1898                          * add this file to the list of results
1899                          * (with each file contributing equal damage).
1900                          */
1901                         damage = 1;
1902                         goto found_damage;
1903                 }
1904
1905                 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
1906                         diff_populate_filespec(p->one, 0);
1907                         diff_populate_filespec(p->two, 0);
1908                         diffcore_count_changes(p->one, p->two, NULL, NULL, 0,
1909                                                &copied, &added);
1910                         diff_free_filespec_data(p->one);
1911                         diff_free_filespec_data(p->two);
1912                 } else if (DIFF_FILE_VALID(p->one)) {
1913                         diff_populate_filespec(p->one, 1);
1914                         copied = added = 0;
1915                         diff_free_filespec_data(p->one);
1916                 } else if (DIFF_FILE_VALID(p->two)) {
1917                         diff_populate_filespec(p->two, 1);
1918                         copied = 0;
1919                         added = p->two->size;
1920                         diff_free_filespec_data(p->two);
1921                 } else
1922                         continue;
1923
1924                 /*
1925                  * Original minus copied is the removed material,
1926                  * added is the new material.  They are both damages
1927                  * made to the preimage.
1928                  * If the resulting damage is zero, we know that
1929                  * diffcore_count_changes() considers the two entries to
1930                  * be identical, but since content_changed is true, we
1931                  * know that there must have been _some_ kind of change,
1932                  * so we force all entries to have damage > 0.
1933                  */
1934                 damage = (p->one->size - copied) + added;
1935                 if (!damage)
1936                         damage = 1;
1937
1938 found_damage:
1939                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1940                 dir.files[dir.nr].name = name;
1941                 dir.files[dir.nr].changed = damage;
1942                 changed += damage;
1943                 dir.nr++;
1944         }
1945
1946         /* This can happen even with many files, if everything was renames */
1947         if (!changed)
1948                 return;
1949
1950         /* Show all directories with more than x% of the changes */
1951         qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
1952         gather_dirstat(options, &dir, changed, "", 0);
1953 }
1954
1955 static void show_dirstat_by_line(struct diffstat_t *data, struct diff_options *options)
1956 {
1957         int i;
1958         unsigned long changed;
1959         struct dirstat_dir dir;
1960
1961         if (data->nr == 0)
1962                 return;
1963
1964         dir.files = NULL;
1965         dir.alloc = 0;
1966         dir.nr = 0;
1967         dir.permille = options->dirstat_permille;
1968         dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
1969
1970         changed = 0;
1971         for (i = 0; i < data->nr; i++) {
1972                 struct diffstat_file *file = data->files[i];
1973                 unsigned long damage = file->added + file->deleted;
1974                 if (file->is_binary)
1975                         /*
1976                          * binary files counts bytes, not lines. Must find some
1977                          * way to normalize binary bytes vs. textual lines.
1978                          * The following heuristic assumes that there are 64
1979                          * bytes per "line".
1980                          * This is stupid and ugly, but very cheap...
1981                          */
1982                         damage = (damage + 63) / 64;
1983                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1984                 dir.files[dir.nr].name = file->name;
1985                 dir.files[dir.nr].changed = damage;
1986                 changed += damage;
1987                 dir.nr++;
1988         }
1989
1990         /* This can happen even with many files, if everything was renames */
1991         if (!changed)
1992                 return;
1993
1994         /* Show all directories with more than x% of the changes */
1995         qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
1996         gather_dirstat(options, &dir, changed, "", 0);
1997 }
1998
1999 static void free_diffstat_info(struct diffstat_t *diffstat)
2000 {
2001         int i;
2002         for (i = 0; i < diffstat->nr; i++) {
2003                 struct diffstat_file *f = diffstat->files[i];
2004                 if (f->name != f->print_name)
2005                         free(f->print_name);
2006                 free(f->name);
2007                 free(f->from_name);
2008                 free(f);
2009         }
2010         free(diffstat->files);
2011 }
2012
2013 struct checkdiff_t {
2014         const char *filename;
2015         int lineno;
2016         int conflict_marker_size;
2017         struct diff_options *o;
2018         unsigned ws_rule;
2019         unsigned status;
2020 };
2021
2022 static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
2023 {
2024         char firstchar;
2025         int cnt;
2026
2027         if (len < marker_size + 1)
2028                 return 0;
2029         firstchar = line[0];
2030         switch (firstchar) {
2031         case '=': case '>': case '<': case '|':
2032                 break;
2033         default:
2034                 return 0;
2035         }
2036         for (cnt = 1; cnt < marker_size; cnt++)
2037                 if (line[cnt] != firstchar)
2038                         return 0;
2039         /* line[1] thru line[marker_size-1] are same as firstchar */
2040         if (len < marker_size + 1 || !isspace(line[marker_size]))
2041                 return 0;
2042         return 1;
2043 }
2044
2045 static void checkdiff_consume(void *priv, char *line, unsigned long len)
2046 {
2047         struct checkdiff_t *data = priv;
2048         int marker_size = data->conflict_marker_size;
2049         const char *ws = diff_get_color(data->o->use_color, DIFF_WHITESPACE);
2050         const char *reset = diff_get_color(data->o->use_color, DIFF_RESET);
2051         const char *set = diff_get_color(data->o->use_color, DIFF_FILE_NEW);
2052         char *err;
2053         const char *line_prefix;
2054
2055         assert(data->o);
2056         line_prefix = diff_line_prefix(data->o);
2057
2058         if (line[0] == '+') {
2059                 unsigned bad;
2060                 data->lineno++;
2061                 if (is_conflict_marker(line + 1, marker_size, len - 1)) {
2062                         data->status |= 1;
2063                         fprintf(data->o->file,
2064                                 "%s%s:%d: leftover conflict marker\n",
2065                                 line_prefix, data->filename, data->lineno);
2066                 }
2067                 bad = ws_check(line + 1, len - 1, data->ws_rule);
2068                 if (!bad)
2069                         return;
2070                 data->status |= bad;
2071                 err = whitespace_error_string(bad);
2072                 fprintf(data->o->file, "%s%s:%d: %s.\n",
2073                         line_prefix, data->filename, data->lineno, err);
2074                 free(err);
2075                 emit_line(data->o, set, reset, line, 1);
2076                 ws_check_emit(line + 1, len - 1, data->ws_rule,
2077                               data->o->file, set, reset, ws);
2078         } else if (line[0] == ' ') {
2079                 data->lineno++;
2080         } else if (line[0] == '@') {
2081                 char *plus = strchr(line, '+');
2082                 if (plus)
2083                         data->lineno = strtol(plus, NULL, 10) - 1;
2084                 else
2085                         die("invalid diff");
2086         }
2087 }
2088
2089 static unsigned char *deflate_it(char *data,
2090                                  unsigned long size,
2091                                  unsigned long *result_size)
2092 {
2093         int bound;
2094         unsigned char *deflated;
2095         git_zstream stream;
2096
2097         memset(&stream, 0, sizeof(stream));
2098         git_deflate_init(&stream, zlib_compression_level);
2099         bound = git_deflate_bound(&stream, size);
2100         deflated = xmalloc(bound);
2101         stream.next_out = deflated;
2102         stream.avail_out = bound;
2103
2104         stream.next_in = (unsigned char *)data;
2105         stream.avail_in = size;
2106         while (git_deflate(&stream, Z_FINISH) == Z_OK)
2107                 ; /* nothing */
2108         git_deflate_end(&stream);
2109         *result_size = stream.total_out;
2110         return deflated;
2111 }
2112
2113 static void emit_binary_diff_body(FILE *file, mmfile_t *one, mmfile_t *two,
2114                                   const char *prefix)
2115 {
2116         void *cp;
2117         void *delta;
2118         void *deflated;
2119         void *data;
2120         unsigned long orig_size;
2121         unsigned long delta_size;
2122         unsigned long deflate_size;
2123         unsigned long data_size;
2124
2125         /* We could do deflated delta, or we could do just deflated two,
2126          * whichever is smaller.
2127          */
2128         delta = NULL;
2129         deflated = deflate_it(two->ptr, two->size, &deflate_size);
2130         if (one->size && two->size) {
2131                 delta = diff_delta(one->ptr, one->size,
2132                                    two->ptr, two->size,
2133                                    &delta_size, deflate_size);
2134                 if (delta) {
2135                         void *to_free = delta;
2136                         orig_size = delta_size;
2137                         delta = deflate_it(delta, delta_size, &delta_size);
2138                         free(to_free);
2139                 }
2140         }
2141
2142         if (delta && delta_size < deflate_size) {
2143                 fprintf(file, "%sdelta %lu\n", prefix, orig_size);
2144                 free(deflated);
2145                 data = delta;
2146                 data_size = delta_size;
2147         }
2148         else {
2149                 fprintf(file, "%sliteral %lu\n", prefix, two->size);
2150                 free(delta);
2151                 data = deflated;
2152                 data_size = deflate_size;
2153         }
2154
2155         /* emit data encoded in base85 */
2156         cp = data;
2157         while (data_size) {
2158                 int bytes = (52 < data_size) ? 52 : data_size;
2159                 char line[70];
2160                 data_size -= bytes;
2161                 if (bytes <= 26)
2162                         line[0] = bytes + 'A' - 1;
2163                 else
2164                         line[0] = bytes - 26 + 'a' - 1;
2165                 encode_85(line + 1, cp, bytes);
2166                 cp = (char *) cp + bytes;
2167                 fprintf(file, "%s", prefix);
2168                 fputs(line, file);
2169                 fputc('\n', file);
2170         }
2171         fprintf(file, "%s\n", prefix);
2172         free(data);
2173 }
2174
2175 static void emit_binary_diff(FILE *file, mmfile_t *one, mmfile_t *two,
2176                              const char *prefix)
2177 {
2178         fprintf(file, "%sGIT binary patch\n", prefix);
2179         emit_binary_diff_body(file, one, two, prefix);
2180         emit_binary_diff_body(file, two, one, prefix);
2181 }
2182
2183 int diff_filespec_is_binary(struct diff_filespec *one)
2184 {
2185         if (one->is_binary == -1) {
2186                 diff_filespec_load_driver(one);
2187                 if (one->driver->binary != -1)
2188                         one->is_binary = one->driver->binary;
2189                 else {
2190                         if (!one->data && DIFF_FILE_VALID(one))
2191                                 diff_populate_filespec(one, 0);
2192                         if (one->data)
2193                                 one->is_binary = buffer_is_binary(one->data,
2194                                                 one->size);
2195                         if (one->is_binary == -1)
2196                                 one->is_binary = 0;
2197                 }
2198         }
2199         return one->is_binary;
2200 }
2201
2202 static const struct userdiff_funcname *diff_funcname_pattern(struct diff_filespec *one)
2203 {
2204         diff_filespec_load_driver(one);
2205         return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
2206 }
2207
2208 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
2209 {
2210         if (!options->a_prefix)
2211                 options->a_prefix = a;
2212         if (!options->b_prefix)
2213                 options->b_prefix = b;
2214 }
2215
2216 struct userdiff_driver *get_textconv(struct diff_filespec *one)
2217 {
2218         if (!DIFF_FILE_VALID(one))
2219                 return NULL;
2220
2221         diff_filespec_load_driver(one);
2222         return userdiff_get_textconv(one->driver);
2223 }
2224
2225 static void builtin_diff(const char *name_a,
2226                          const char *name_b,
2227                          struct diff_filespec *one,
2228                          struct diff_filespec *two,
2229                          const char *xfrm_msg,
2230                          int must_show_header,
2231                          struct diff_options *o,
2232                          int complete_rewrite)
2233 {
2234         mmfile_t mf1, mf2;
2235         const char *lbl[2];
2236         char *a_one, *b_two;
2237         const char *meta = diff_get_color_opt(o, DIFF_METAINFO);
2238         const char *reset = diff_get_color_opt(o, DIFF_RESET);
2239         const char *a_prefix, *b_prefix;
2240         struct userdiff_driver *textconv_one = NULL;
2241         struct userdiff_driver *textconv_two = NULL;
2242         struct strbuf header = STRBUF_INIT;
2243         const char *line_prefix = diff_line_prefix(o);
2244
2245         if (DIFF_OPT_TST(o, SUBMODULE_LOG) &&
2246                         (!one->mode || S_ISGITLINK(one->mode)) &&
2247                         (!two->mode || S_ISGITLINK(two->mode))) {
2248                 const char *del = diff_get_color_opt(o, DIFF_FILE_OLD);
2249                 const char *add = diff_get_color_opt(o, DIFF_FILE_NEW);
2250                 show_submodule_summary(o->file, one->path ? one->path : two->path,
2251                                 line_prefix,
2252                                 one->sha1, two->sha1, two->dirty_submodule,
2253                                 meta, del, add, reset);
2254                 return;
2255         }
2256
2257         if (DIFF_OPT_TST(o, ALLOW_TEXTCONV)) {
2258                 textconv_one = get_textconv(one);
2259                 textconv_two = get_textconv(two);
2260         }
2261
2262         diff_set_mnemonic_prefix(o, "a/", "b/");
2263         if (DIFF_OPT_TST(o, REVERSE_DIFF)) {
2264                 a_prefix = o->b_prefix;
2265                 b_prefix = o->a_prefix;
2266         } else {
2267                 a_prefix = o->a_prefix;
2268                 b_prefix = o->b_prefix;
2269         }
2270
2271         /* Never use a non-valid filename anywhere if at all possible */
2272         name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
2273         name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
2274
2275         a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
2276         b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
2277         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
2278         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
2279         strbuf_addf(&header, "%s%sdiff --git %s %s%s\n", line_prefix, meta, a_one, b_two, reset);
2280         if (lbl[0][0] == '/') {
2281                 /* /dev/null */
2282                 strbuf_addf(&header, "%s%snew file mode %06o%s\n", line_prefix, meta, two->mode, reset);
2283                 if (xfrm_msg)
2284                         strbuf_addstr(&header, xfrm_msg);
2285                 must_show_header = 1;
2286         }
2287         else if (lbl[1][0] == '/') {
2288                 strbuf_addf(&header, "%s%sdeleted file mode %06o%s\n", line_prefix, meta, one->mode, reset);
2289                 if (xfrm_msg)
2290                         strbuf_addstr(&header, xfrm_msg);
2291                 must_show_header = 1;
2292         }
2293         else {
2294                 if (one->mode != two->mode) {
2295                         strbuf_addf(&header, "%s%sold mode %06o%s\n", line_prefix, meta, one->mode, reset);
2296                         strbuf_addf(&header, "%s%snew mode %06o%s\n", line_prefix, meta, two->mode, reset);
2297                         must_show_header = 1;
2298                 }
2299                 if (xfrm_msg)
2300                         strbuf_addstr(&header, xfrm_msg);
2301
2302                 /*
2303                  * we do not run diff between different kind
2304                  * of objects.
2305                  */
2306                 if ((one->mode ^ two->mode) & S_IFMT)
2307                         goto free_ab_and_return;
2308                 if (complete_rewrite &&
2309                     (textconv_one || !diff_filespec_is_binary(one)) &&
2310                     (textconv_two || !diff_filespec_is_binary(two))) {
2311                         fprintf(o->file, "%s", header.buf);
2312                         strbuf_reset(&header);
2313                         emit_rewrite_diff(name_a, name_b, one, two,
2314                                                 textconv_one, textconv_two, o);
2315                         o->found_changes = 1;
2316                         goto free_ab_and_return;
2317                 }
2318         }
2319
2320         if (o->irreversible_delete && lbl[1][0] == '/') {
2321                 fprintf(o->file, "%s", header.buf);
2322                 strbuf_reset(&header);
2323                 goto free_ab_and_return;
2324         } else if (!DIFF_OPT_TST(o, TEXT) &&
2325             ( (!textconv_one && diff_filespec_is_binary(one)) ||
2326               (!textconv_two && diff_filespec_is_binary(two)) )) {
2327                 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2328                         die("unable to read files to diff");
2329                 /* Quite common confusing case */
2330                 if (mf1.size == mf2.size &&
2331                     !memcmp(mf1.ptr, mf2.ptr, mf1.size)) {
2332                         if (must_show_header)
2333                                 fprintf(o->file, "%s", header.buf);
2334                         goto free_ab_and_return;
2335                 }
2336                 fprintf(o->file, "%s", header.buf);
2337                 strbuf_reset(&header);
2338                 if (DIFF_OPT_TST(o, BINARY))
2339                         emit_binary_diff(o->file, &mf1, &mf2, line_prefix);
2340                 else
2341                         fprintf(o->file, "%sBinary files %s and %s differ\n",
2342                                 line_prefix, lbl[0], lbl[1]);
2343                 o->found_changes = 1;
2344         } else {
2345                 /* Crazy xdl interfaces.. */
2346                 const char *diffopts = getenv("GIT_DIFF_OPTS");
2347                 const char *v;
2348                 xpparam_t xpp;
2349                 xdemitconf_t xecfg;
2350                 struct emit_callback ecbdata;
2351                 const struct userdiff_funcname *pe;
2352
2353                 if (must_show_header) {
2354                         fprintf(o->file, "%s", header.buf);
2355                         strbuf_reset(&header);
2356                 }
2357
2358                 mf1.size = fill_textconv(textconv_one, one, &mf1.ptr);
2359                 mf2.size = fill_textconv(textconv_two, two, &mf2.ptr);
2360
2361                 pe = diff_funcname_pattern(one);
2362                 if (!pe)
2363                         pe = diff_funcname_pattern(two);
2364
2365                 memset(&xpp, 0, sizeof(xpp));
2366                 memset(&xecfg, 0, sizeof(xecfg));
2367                 memset(&ecbdata, 0, sizeof(ecbdata));
2368                 ecbdata.label_path = lbl;
2369                 ecbdata.color_diff = want_color(o->use_color);
2370                 ecbdata.found_changesp = &o->found_changes;
2371                 ecbdata.ws_rule = whitespace_rule(name_b);
2372                 if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
2373                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
2374                 ecbdata.opt = o;
2375                 ecbdata.header = header.len ? &header : NULL;
2376                 xpp.flags = o->xdl_opts;
2377                 xecfg.ctxlen = o->context;
2378                 xecfg.interhunkctxlen = o->interhunkcontext;
2379                 xecfg.flags = XDL_EMIT_FUNCNAMES;
2380                 if (DIFF_OPT_TST(o, FUNCCONTEXT))
2381                         xecfg.flags |= XDL_EMIT_FUNCCONTEXT;
2382                 if (pe)
2383                         xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
2384                 if (!diffopts)
2385                         ;
2386                 else if (skip_prefix(diffopts, "--unified=", &v))
2387                         xecfg.ctxlen = strtoul(v, NULL, 10);
2388                 else if (skip_prefix(diffopts, "-u", &v))
2389                         xecfg.ctxlen = strtoul(v, NULL, 10);
2390                 if (o->word_diff)
2391                         init_diff_words_data(&ecbdata, o, one, two);
2392                 xdi_diff_outf(&mf1, &mf2, fn_out_consume, &ecbdata,
2393                               &xpp, &xecfg);
2394                 if (o->word_diff)
2395                         free_diff_words_data(&ecbdata);
2396                 if (textconv_one)
2397                         free(mf1.ptr);
2398                 if (textconv_two)
2399                         free(mf2.ptr);
2400                 xdiff_clear_find_func(&xecfg);
2401         }
2402
2403  free_ab_and_return:
2404         strbuf_release(&header);
2405         diff_free_filespec_data(one);
2406         diff_free_filespec_data(two);
2407         free(a_one);
2408         free(b_two);
2409         return;
2410 }
2411
2412 static void builtin_diffstat(const char *name_a, const char *name_b,
2413                              struct diff_filespec *one,
2414                              struct diff_filespec *two,
2415                              struct diffstat_t *diffstat,
2416                              struct diff_options *o,
2417                              struct diff_filepair *p)
2418 {
2419         mmfile_t mf1, mf2;
2420         struct diffstat_file *data;
2421         int same_contents;
2422         int complete_rewrite = 0;
2423
2424         if (!DIFF_PAIR_UNMERGED(p)) {
2425                 if (p->status == DIFF_STATUS_MODIFIED && p->score)
2426                         complete_rewrite = 1;
2427         }
2428
2429         data = diffstat_add(diffstat, name_a, name_b);
2430         data->is_interesting = p->status != DIFF_STATUS_UNKNOWN;
2431
2432         if (!one || !two) {
2433                 data->is_unmerged = 1;
2434                 return;
2435         }
2436
2437         same_contents = !hashcmp(one->sha1, two->sha1);
2438
2439         if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
2440                 data->is_binary = 1;
2441                 if (same_contents) {
2442                         data->added = 0;
2443                         data->deleted = 0;
2444                 } else {
2445                         data->added = diff_filespec_size(two);
2446                         data->deleted = diff_filespec_size(one);
2447                 }
2448         }
2449
2450         else if (complete_rewrite) {
2451                 diff_populate_filespec(one, 0);
2452                 diff_populate_filespec(two, 0);
2453                 data->deleted = count_lines(one->data, one->size);
2454                 data->added = count_lines(two->data, two->size);
2455         }
2456
2457         else if (!same_contents) {
2458                 /* Crazy xdl interfaces.. */
2459                 xpparam_t xpp;
2460                 xdemitconf_t xecfg;
2461
2462                 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2463                         die("unable to read files to diff");
2464
2465                 memset(&xpp, 0, sizeof(xpp));
2466                 memset(&xecfg, 0, sizeof(xecfg));
2467                 xpp.flags = o->xdl_opts;
2468                 xecfg.ctxlen = o->context;
2469                 xecfg.interhunkctxlen = o->interhunkcontext;
2470                 xdi_diff_outf(&mf1, &mf2, diffstat_consume, diffstat,
2471                               &xpp, &xecfg);
2472         }
2473
2474         diff_free_filespec_data(one);
2475         diff_free_filespec_data(two);
2476 }
2477
2478 static void builtin_checkdiff(const char *name_a, const char *name_b,
2479                               const char *attr_path,
2480                               struct diff_filespec *one,
2481                               struct diff_filespec *two,
2482                               struct diff_options *o)
2483 {
2484         mmfile_t mf1, mf2;
2485         struct checkdiff_t data;
2486
2487         if (!two)
2488                 return;
2489
2490         memset(&data, 0, sizeof(data));
2491         data.filename = name_b ? name_b : name_a;
2492         data.lineno = 0;
2493         data.o = o;
2494         data.ws_rule = whitespace_rule(attr_path);
2495         data.conflict_marker_size = ll_merge_marker_size(attr_path);
2496
2497         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2498                 die("unable to read files to diff");
2499
2500         /*
2501          * All the other codepaths check both sides, but not checking
2502          * the "old" side here is deliberate.  We are checking the newly
2503          * introduced changes, and as long as the "new" side is text, we
2504          * can and should check what it introduces.
2505          */
2506         if (diff_filespec_is_binary(two))
2507                 goto free_and_return;
2508         else {
2509                 /* Crazy xdl interfaces.. */
2510                 xpparam_t xpp;
2511                 xdemitconf_t xecfg;
2512
2513                 memset(&xpp, 0, sizeof(xpp));
2514                 memset(&xecfg, 0, sizeof(xecfg));
2515                 xecfg.ctxlen = 1; /* at least one context line */
2516                 xpp.flags = 0;
2517                 xdi_diff_outf(&mf1, &mf2, checkdiff_consume, &data,
2518                               &xpp, &xecfg);
2519
2520                 if (data.ws_rule & WS_BLANK_AT_EOF) {
2521                         struct emit_callback ecbdata;
2522                         int blank_at_eof;
2523
2524                         ecbdata.ws_rule = data.ws_rule;
2525                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
2526                         blank_at_eof = ecbdata.blank_at_eof_in_postimage;
2527
2528                         if (blank_at_eof) {
2529                                 static char *err;
2530                                 if (!err)
2531                                         err = whitespace_error_string(WS_BLANK_AT_EOF);
2532                                 fprintf(o->file, "%s:%d: %s.\n",
2533                                         data.filename, blank_at_eof, err);
2534                                 data.status = 1; /* report errors */
2535                         }
2536                 }
2537         }
2538  free_and_return:
2539         diff_free_filespec_data(one);
2540         diff_free_filespec_data(two);
2541         if (data.status)
2542                 DIFF_OPT_SET(o, CHECK_FAILED);
2543 }
2544
2545 struct diff_filespec *alloc_filespec(const char *path)
2546 {
2547         int namelen = strlen(path);
2548         struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
2549
2550         memset(spec, 0, sizeof(*spec));
2551         spec->path = (char *)(spec + 1);
2552         memcpy(spec->path, path, namelen+1);
2553         spec->count = 1;
2554         spec->is_binary = -1;
2555         return spec;
2556 }
2557
2558 void free_filespec(struct diff_filespec *spec)
2559 {
2560         if (!--spec->count) {
2561                 diff_free_filespec_data(spec);
2562                 free(spec);
2563         }
2564 }
2565
2566 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
2567                    int sha1_valid, unsigned short mode)
2568 {
2569         if (mode) {
2570                 spec->mode = canon_mode(mode);
2571                 hashcpy(spec->sha1, sha1);
2572                 spec->sha1_valid = sha1_valid;
2573         }
2574 }
2575
2576 /*
2577  * Given a name and sha1 pair, if the index tells us the file in
2578  * the work tree has that object contents, return true, so that
2579  * prepare_temp_file() does not have to inflate and extract.
2580  */
2581 static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
2582 {
2583         const struct cache_entry *ce;
2584         struct stat st;
2585         int pos, len;
2586
2587         /*
2588          * We do not read the cache ourselves here, because the
2589          * benchmark with my previous version that always reads cache
2590          * shows that it makes things worse for diff-tree comparing
2591          * two linux-2.6 kernel trees in an already checked out work
2592          * tree.  This is because most diff-tree comparisons deal with
2593          * only a small number of files, while reading the cache is
2594          * expensive for a large project, and its cost outweighs the
2595          * savings we get by not inflating the object to a temporary
2596          * file.  Practically, this code only helps when we are used
2597          * by diff-cache --cached, which does read the cache before
2598          * calling us.
2599          */
2600         if (!active_cache)
2601                 return 0;
2602
2603         /* We want to avoid the working directory if our caller
2604          * doesn't need the data in a normal file, this system
2605          * is rather slow with its stat/open/mmap/close syscalls,
2606          * and the object is contained in a pack file.  The pack
2607          * is probably already open and will be faster to obtain
2608          * the data through than the working directory.  Loose
2609          * objects however would tend to be slower as they need
2610          * to be individually opened and inflated.
2611          */
2612         if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1))
2613                 return 0;
2614
2615         len = strlen(name);
2616         pos = cache_name_pos(name, len);
2617         if (pos < 0)
2618                 return 0;
2619         ce = active_cache[pos];
2620
2621         /*
2622          * This is not the sha1 we are looking for, or
2623          * unreusable because it is not a regular file.
2624          */
2625         if (hashcmp(sha1, ce->sha1) || !S_ISREG(ce->ce_mode))
2626                 return 0;
2627
2628         /*
2629          * If ce is marked as "assume unchanged", there is no
2630          * guarantee that work tree matches what we are looking for.
2631          */
2632         if ((ce->ce_flags & CE_VALID) || ce_skip_worktree(ce))
2633                 return 0;
2634
2635         /*
2636          * If ce matches the file in the work tree, we can reuse it.
2637          */
2638         if (ce_uptodate(ce) ||
2639             (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
2640                 return 1;
2641
2642         return 0;
2643 }
2644
2645 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
2646 {
2647         int len;
2648         char *data = xmalloc(100), *dirty = "";
2649
2650         /* Are we looking at the work tree? */
2651         if (s->dirty_submodule)
2652                 dirty = "-dirty";
2653
2654         len = snprintf(data, 100,
2655                        "Subproject commit %s%s\n", sha1_to_hex(s->sha1), dirty);
2656         s->data = data;
2657         s->size = len;
2658         s->should_free = 1;
2659         if (size_only) {
2660                 s->data = NULL;
2661                 free(data);
2662         }
2663         return 0;
2664 }
2665
2666 /*
2667  * While doing rename detection and pickaxe operation, we may need to
2668  * grab the data for the blob (or file) for our own in-core comparison.
2669  * diff_filespec has data and size fields for this purpose.
2670  */
2671 int diff_populate_filespec(struct diff_filespec *s, int size_only)
2672 {
2673         int err = 0;
2674         /*
2675          * demote FAIL to WARN to allow inspecting the situation
2676          * instead of refusing.
2677          */
2678         enum safe_crlf crlf_warn = (safe_crlf == SAFE_CRLF_FAIL
2679                                     ? SAFE_CRLF_WARN
2680                                     : safe_crlf);
2681
2682         if (!DIFF_FILE_VALID(s))
2683                 die("internal error: asking to populate invalid file.");
2684         if (S_ISDIR(s->mode))
2685                 return -1;
2686
2687         if (s->data)
2688                 return 0;
2689
2690         if (size_only && 0 < s->size)
2691                 return 0;
2692
2693         if (S_ISGITLINK(s->mode))
2694                 return diff_populate_gitlink(s, size_only);
2695
2696         if (!s->sha1_valid ||
2697             reuse_worktree_file(s->path, s->sha1, 0)) {
2698                 struct strbuf buf = STRBUF_INIT;
2699                 struct stat st;
2700                 int fd;
2701
2702                 if (lstat(s->path, &st) < 0) {
2703                         if (errno == ENOENT) {
2704                         err_empty:
2705                                 err = -1;
2706                         empty:
2707                                 s->data = (char *)"";
2708                                 s->size = 0;
2709                                 return err;
2710                         }
2711                 }
2712                 s->size = xsize_t(st.st_size);
2713                 if (!s->size)
2714                         goto empty;
2715                 if (S_ISLNK(st.st_mode)) {
2716                         struct strbuf sb = STRBUF_INIT;
2717
2718                         if (strbuf_readlink(&sb, s->path, s->size))
2719                                 goto err_empty;
2720                         s->size = sb.len;
2721                         s->data = strbuf_detach(&sb, NULL);
2722                         s->should_free = 1;
2723                         return 0;
2724                 }
2725                 if (size_only)
2726                         return 0;
2727                 fd = open(s->path, O_RDONLY);
2728                 if (fd < 0)
2729                         goto err_empty;
2730                 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
2731                 close(fd);
2732                 s->should_munmap = 1;
2733
2734                 /*
2735                  * Convert from working tree format to canonical git format
2736                  */
2737                 if (convert_to_git(s->path, s->data, s->size, &buf, crlf_warn)) {
2738                         size_t size = 0;
2739                         munmap(s->data, s->size);
2740                         s->should_munmap = 0;
2741                         s->data = strbuf_detach(&buf, &size);
2742                         s->size = size;
2743                         s->should_free = 1;
2744                 }
2745         }
2746         else {
2747                 enum object_type type;
2748                 if (size_only) {
2749                         type = sha1_object_info(s->sha1, &s->size);
2750                         if (type < 0)
2751                                 die("unable to read %s", sha1_to_hex(s->sha1));
2752                 } else {
2753                         s->data = read_sha1_file(s->sha1, &type, &s->size);
2754                         if (!s->data)
2755                                 die("unable to read %s", sha1_to_hex(s->sha1));
2756                         s->should_free = 1;
2757                 }
2758         }
2759         return 0;
2760 }
2761
2762 void diff_free_filespec_blob(struct diff_filespec *s)
2763 {
2764         if (s->should_free)
2765                 free(s->data);
2766         else if (s->should_munmap)
2767                 munmap(s->data, s->size);
2768
2769         if (s->should_free || s->should_munmap) {
2770                 s->should_free = s->should_munmap = 0;
2771                 s->data = NULL;
2772         }
2773 }
2774
2775 void diff_free_filespec_data(struct diff_filespec *s)
2776 {
2777         diff_free_filespec_blob(s);
2778         free(s->cnt_data);
2779         s->cnt_data = NULL;
2780 }
2781
2782 static void prep_temp_blob(const char *path, struct diff_tempfile *temp,
2783                            void *blob,
2784                            unsigned long size,
2785                            const unsigned char *sha1,
2786                            int mode)
2787 {
2788         int fd;
2789         struct strbuf buf = STRBUF_INIT;
2790         struct strbuf template = STRBUF_INIT;
2791         char *path_dup = xstrdup(path);
2792         const char *base = basename(path_dup);
2793
2794         /* Generate "XXXXXX_basename.ext" */
2795         strbuf_addstr(&template, "XXXXXX_");
2796         strbuf_addstr(&template, base);
2797
2798         fd = git_mkstemps(temp->tmp_path, PATH_MAX, template.buf,
2799                         strlen(base) + 1);
2800         if (fd < 0)
2801                 die_errno("unable to create temp-file");
2802         if (convert_to_working_tree(path,
2803                         (const char *)blob, (size_t)size, &buf)) {
2804                 blob = buf.buf;
2805                 size = buf.len;
2806         }
2807         if (write_in_full(fd, blob, size) != size)
2808                 die_errno("unable to write temp-file");
2809         close(fd);
2810         temp->name = temp->tmp_path;
2811         strcpy(temp->hex, sha1_to_hex(sha1));
2812         temp->hex[40] = 0;
2813         sprintf(temp->mode, "%06o", mode);
2814         strbuf_release(&buf);
2815         strbuf_release(&template);
2816         free(path_dup);
2817 }
2818
2819 static struct diff_tempfile *prepare_temp_file(const char *name,
2820                 struct diff_filespec *one)
2821 {
2822         struct diff_tempfile *temp = claim_diff_tempfile();
2823
2824         if (!DIFF_FILE_VALID(one)) {
2825         not_a_valid_file:
2826                 /* A '-' entry produces this for file-2, and
2827                  * a '+' entry produces this for file-1.
2828                  */
2829                 temp->name = "/dev/null";
2830                 strcpy(temp->hex, ".");
2831                 strcpy(temp->mode, ".");
2832                 return temp;
2833         }
2834
2835         if (!remove_tempfile_installed) {
2836                 atexit(remove_tempfile);
2837                 sigchain_push_common(remove_tempfile_on_signal);
2838                 remove_tempfile_installed = 1;
2839         }
2840
2841         if (!S_ISGITLINK(one->mode) &&
2842             (!one->sha1_valid ||
2843              reuse_worktree_file(name, one->sha1, 1))) {
2844                 struct stat st;
2845                 if (lstat(name, &st) < 0) {
2846                         if (errno == ENOENT)
2847                                 goto not_a_valid_file;
2848                         die_errno("stat(%s)", name);
2849                 }
2850                 if (S_ISLNK(st.st_mode)) {
2851                         struct strbuf sb = STRBUF_INIT;
2852                         if (strbuf_readlink(&sb, name, st.st_size) < 0)
2853                                 die_errno("readlink(%s)", name);
2854                         prep_temp_blob(name, temp, sb.buf, sb.len,
2855                                        (one->sha1_valid ?
2856                                         one->sha1 : null_sha1),
2857                                        (one->sha1_valid ?
2858                                         one->mode : S_IFLNK));
2859                         strbuf_release(&sb);
2860                 }
2861                 else {
2862                         /* we can borrow from the file in the work tree */
2863                         temp->name = name;
2864                         if (!one->sha1_valid)
2865                                 strcpy(temp->hex, sha1_to_hex(null_sha1));
2866                         else
2867                                 strcpy(temp->hex, sha1_to_hex(one->sha1));
2868                         /* Even though we may sometimes borrow the
2869                          * contents from the work tree, we always want
2870                          * one->mode.  mode is trustworthy even when
2871                          * !(one->sha1_valid), as long as
2872                          * DIFF_FILE_VALID(one).
2873                          */
2874                         sprintf(temp->mode, "%06o", one->mode);
2875                 }
2876                 return temp;
2877         }
2878         else {
2879                 if (diff_populate_filespec(one, 0))
2880                         die("cannot read data blob for %s", one->path);
2881                 prep_temp_blob(name, temp, one->data, one->size,
2882                                one->sha1, one->mode);
2883         }
2884         return temp;
2885 }
2886
2887 static void add_external_diff_name(struct argv_array *argv,
2888                                    const char *name,
2889                                    struct diff_filespec *df)
2890 {
2891         struct diff_tempfile *temp = prepare_temp_file(name, df);
2892         argv_array_push(argv, temp->name);
2893         argv_array_push(argv, temp->hex);
2894         argv_array_push(argv, temp->mode);
2895 }
2896
2897 /* An external diff command takes:
2898  *
2899  * diff-cmd name infile1 infile1-sha1 infile1-mode \
2900  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
2901  *
2902  */
2903 static void run_external_diff(const char *pgm,
2904                               const char *name,
2905                               const char *other,
2906                               struct diff_filespec *one,
2907                               struct diff_filespec *two,
2908                               const char *xfrm_msg,
2909                               int complete_rewrite,
2910                               struct diff_options *o)
2911 {
2912         struct argv_array argv = ARGV_ARRAY_INIT;
2913         struct argv_array env = ARGV_ARRAY_INIT;
2914         struct diff_queue_struct *q = &diff_queued_diff;
2915
2916         argv_array_push(&argv, pgm);
2917         argv_array_push(&argv, name);
2918
2919         if (one && two) {
2920                 add_external_diff_name(&argv, name, one);
2921                 if (!other)
2922                         add_external_diff_name(&argv, name, two);
2923                 else {
2924                         add_external_diff_name(&argv, other, two);
2925                         argv_array_push(&argv, other);
2926                         argv_array_push(&argv, xfrm_msg);
2927                 }
2928         }
2929
2930         argv_array_pushf(&env, "GIT_DIFF_PATH_COUNTER=%d", ++o->diff_path_counter);
2931         argv_array_pushf(&env, "GIT_DIFF_PATH_TOTAL=%d", q->nr);
2932
2933         if (run_command_v_opt_cd_env(argv.argv, RUN_USING_SHELL, NULL, env.argv))
2934                 die(_("external diff died, stopping at %s"), name);
2935
2936         remove_tempfile();
2937         argv_array_clear(&argv);
2938         argv_array_clear(&env);
2939 }
2940
2941 static int similarity_index(struct diff_filepair *p)
2942 {
2943         return p->score * 100 / MAX_SCORE;
2944 }
2945
2946 static void fill_metainfo(struct strbuf *msg,
2947                           const char *name,
2948                           const char *other,
2949                           struct diff_filespec *one,
2950                           struct diff_filespec *two,
2951                           struct diff_options *o,
2952                           struct diff_filepair *p,
2953                           int *must_show_header,
2954                           int use_color)
2955 {
2956         const char *set = diff_get_color(use_color, DIFF_METAINFO);
2957         const char *reset = diff_get_color(use_color, DIFF_RESET);
2958         const char *line_prefix = diff_line_prefix(o);
2959
2960         *must_show_header = 1;
2961         strbuf_init(msg, PATH_MAX * 2 + 300);
2962         switch (p->status) {
2963         case DIFF_STATUS_COPIED:
2964                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
2965                             line_prefix, set, similarity_index(p));
2966                 strbuf_addf(msg, "%s\n%s%scopy from ",
2967                             reset,  line_prefix, set);
2968                 quote_c_style(name, msg, NULL, 0);
2969                 strbuf_addf(msg, "%s\n%s%scopy to ", reset, line_prefix, set);
2970                 quote_c_style(other, msg, NULL, 0);
2971                 strbuf_addf(msg, "%s\n", reset);
2972                 break;
2973         case DIFF_STATUS_RENAMED:
2974                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
2975                             line_prefix, set, similarity_index(p));
2976                 strbuf_addf(msg, "%s\n%s%srename from ",
2977                             reset, line_prefix, set);
2978                 quote_c_style(name, msg, NULL, 0);
2979                 strbuf_addf(msg, "%s\n%s%srename to ",
2980                             reset, line_prefix, set);
2981                 quote_c_style(other, msg, NULL, 0);
2982                 strbuf_addf(msg, "%s\n", reset);
2983                 break;
2984         case DIFF_STATUS_MODIFIED:
2985                 if (p->score) {
2986                         strbuf_addf(msg, "%s%sdissimilarity index %d%%%s\n",
2987                                     line_prefix,
2988                                     set, similarity_index(p), reset);
2989                         break;
2990                 }
2991                 /* fallthru */
2992         default:
2993                 *must_show_header = 0;
2994         }
2995         if (one && two && hashcmp(one->sha1, two->sha1)) {
2996                 int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
2997
2998                 if (DIFF_OPT_TST(o, BINARY)) {
2999                         mmfile_t mf;
3000                         if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
3001                             (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
3002                                 abbrev = 40;
3003                 }
3004                 strbuf_addf(msg, "%s%sindex %s..", line_prefix, set,
3005                             find_unique_abbrev(one->sha1, abbrev));
3006                 strbuf_addstr(msg, find_unique_abbrev(two->sha1, abbrev));
3007                 if (one->mode == two->mode)
3008                         strbuf_addf(msg, " %06o", one->mode);
3009                 strbuf_addf(msg, "%s\n", reset);
3010         }
3011 }
3012
3013 static void run_diff_cmd(const char *pgm,
3014                          const char *name,
3015                          const char *other,
3016                          const char *attr_path,
3017                          struct diff_filespec *one,
3018                          struct diff_filespec *two,
3019                          struct strbuf *msg,
3020                          struct diff_options *o,
3021                          struct diff_filepair *p)
3022 {
3023         const char *xfrm_msg = NULL;
3024         int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
3025         int must_show_header = 0;
3026
3027
3028         if (DIFF_OPT_TST(o, ALLOW_EXTERNAL)) {
3029                 struct userdiff_driver *drv = userdiff_find_by_path(attr_path);
3030                 if (drv && drv->external)
3031                         pgm = drv->external;
3032         }
3033
3034         if (msg) {
3035                 /*
3036                  * don't use colors when the header is intended for an
3037                  * external diff driver
3038                  */
3039                 fill_metainfo(msg, name, other, one, two, o, p,
3040                               &must_show_header,
3041                               want_color(o->use_color) && !pgm);
3042                 xfrm_msg = msg->len ? msg->buf : NULL;
3043         }
3044
3045         if (pgm) {
3046                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
3047                                   complete_rewrite, o);
3048                 return;
3049         }
3050         if (one && two)
3051                 builtin_diff(name, other ? other : name,
3052                              one, two, xfrm_msg, must_show_header,
3053                              o, complete_rewrite);
3054         else
3055                 fprintf(o->file, "* Unmerged path %s\n", name);
3056 }
3057
3058 static void diff_fill_sha1_info(struct diff_filespec *one)
3059 {
3060         if (DIFF_FILE_VALID(one)) {
3061                 if (!one->sha1_valid) {
3062                         struct stat st;
3063                         if (one->is_stdin) {
3064                                 hashcpy(one->sha1, null_sha1);
3065                                 return;
3066                         }
3067                         if (lstat(one->path, &st) < 0)
3068                                 die_errno("stat '%s'", one->path);
3069                         if (index_path(one->sha1, one->path, &st, 0))
3070                                 die("cannot hash %s", one->path);
3071                 }
3072         }
3073         else
3074                 hashclr(one->sha1);
3075 }
3076
3077 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
3078 {
3079         /* Strip the prefix but do not molest /dev/null and absolute paths */
3080         if (*namep && **namep != '/') {
3081                 *namep += prefix_length;
3082                 if (**namep == '/')
3083                         ++*namep;
3084         }
3085         if (*otherp && **otherp != '/') {
3086                 *otherp += prefix_length;
3087                 if (**otherp == '/')
3088                         ++*otherp;
3089         }
3090 }
3091
3092 static void run_diff(struct diff_filepair *p, struct diff_options *o)
3093 {
3094         const char *pgm = external_diff();
3095         struct strbuf msg;
3096         struct diff_filespec *one = p->one;
3097         struct diff_filespec *two = p->two;
3098         const char *name;
3099         const char *other;
3100         const char *attr_path;
3101
3102         name  = p->one->path;
3103         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3104         attr_path = name;
3105         if (o->prefix_length)
3106                 strip_prefix(o->prefix_length, &name, &other);
3107
3108         if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
3109                 pgm = NULL;
3110
3111         if (DIFF_PAIR_UNMERGED(p)) {
3112                 run_diff_cmd(pgm, name, NULL, attr_path,
3113                              NULL, NULL, NULL, o, p);
3114                 return;
3115         }
3116
3117         diff_fill_sha1_info(one);
3118         diff_fill_sha1_info(two);
3119
3120         if (!pgm &&
3121             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
3122             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
3123                 /*
3124                  * a filepair that changes between file and symlink
3125                  * needs to be split into deletion and creation.
3126                  */
3127                 struct diff_filespec *null = alloc_filespec(two->path);
3128                 run_diff_cmd(NULL, name, other, attr_path,
3129                              one, null, &msg, o, p);
3130                 free(null);
3131                 strbuf_release(&msg);
3132
3133                 null = alloc_filespec(one->path);
3134                 run_diff_cmd(NULL, name, other, attr_path,
3135                              null, two, &msg, o, p);
3136                 free(null);
3137         }
3138         else
3139                 run_diff_cmd(pgm, name, other, attr_path,
3140                              one, two, &msg, o, p);
3141
3142         strbuf_release(&msg);
3143 }
3144
3145 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
3146                          struct diffstat_t *diffstat)
3147 {
3148         const char *name;
3149         const char *other;
3150
3151         if (DIFF_PAIR_UNMERGED(p)) {
3152                 /* unmerged */
3153                 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, p);
3154                 return;
3155         }
3156
3157         name = p->one->path;
3158         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3159
3160         if (o->prefix_length)
3161                 strip_prefix(o->prefix_length, &name, &other);
3162
3163         diff_fill_sha1_info(p->one);
3164         diff_fill_sha1_info(p->two);
3165
3166         builtin_diffstat(name, other, p->one, p->two, diffstat, o, p);
3167 }
3168
3169 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
3170 {
3171         const char *name;
3172         const char *other;
3173         const char *attr_path;
3174
3175         if (DIFF_PAIR_UNMERGED(p)) {
3176                 /* unmerged */
3177                 return;
3178         }
3179
3180         name = p->one->path;
3181         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3182         attr_path = other ? other : name;
3183
3184         if (o->prefix_length)
3185                 strip_prefix(o->prefix_length, &name, &other);
3186
3187         diff_fill_sha1_info(p->one);
3188         diff_fill_sha1_info(p->two);
3189
3190         builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
3191 }
3192
3193 void diff_setup(struct diff_options *options)
3194 {
3195         memcpy(options, &default_diff_options, sizeof(*options));
3196
3197         options->file = stdout;
3198
3199         options->line_termination = '\n';
3200         options->break_opt = -1;
3201         options->rename_limit = -1;
3202         options->dirstat_permille = diff_dirstat_permille_default;
3203         options->context = diff_context_default;
3204         DIFF_OPT_SET(options, RENAME_EMPTY);
3205
3206         /* pathchange left =NULL by default */
3207         options->change = diff_change;
3208         options->add_remove = diff_addremove;
3209         options->use_color = diff_use_color_default;
3210         options->detect_rename = diff_detect_rename_default;
3211         options->xdl_opts |= diff_algorithm;
3212
3213         options->orderfile = diff_order_file_cfg;
3214
3215         if (diff_no_prefix) {
3216                 options->a_prefix = options->b_prefix = "";
3217         } else if (!diff_mnemonic_prefix) {
3218                 options->a_prefix = "a/";
3219                 options->b_prefix = "b/";
3220         }
3221 }
3222
3223 void diff_setup_done(struct diff_options *options)
3224 {
3225         int count = 0;
3226
3227         if (options->set_default)
3228                 options->set_default(options);
3229
3230         if (options->output_format & DIFF_FORMAT_NAME)
3231                 count++;
3232         if (options->output_format & DIFF_FORMAT_NAME_STATUS)
3233                 count++;
3234         if (options->output_format & DIFF_FORMAT_CHECKDIFF)
3235                 count++;
3236         if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
3237                 count++;
3238         if (count > 1)
3239                 die("--name-only, --name-status, --check and -s are mutually exclusive");
3240
3241         /*
3242          * Most of the time we can say "there are changes"
3243          * only by checking if there are changed paths, but
3244          * --ignore-whitespace* options force us to look
3245          * inside contents.
3246          */
3247
3248         if (DIFF_XDL_TST(options, IGNORE_WHITESPACE) ||
3249             DIFF_XDL_TST(options, IGNORE_WHITESPACE_CHANGE) ||
3250             DIFF_XDL_TST(options, IGNORE_WHITESPACE_AT_EOL))
3251                 DIFF_OPT_SET(options, DIFF_FROM_CONTENTS);
3252         else
3253                 DIFF_OPT_CLR(options, DIFF_FROM_CONTENTS);
3254
3255         if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
3256                 options->detect_rename = DIFF_DETECT_COPY;
3257
3258         if (!DIFF_OPT_TST(options, RELATIVE_NAME))
3259                 options->prefix = NULL;
3260         if (options->prefix)
3261                 options->prefix_length = strlen(options->prefix);
3262         else
3263                 options->prefix_length = 0;
3264
3265         if (options->output_format & (DIFF_FORMAT_NAME |
3266                                       DIFF_FORMAT_NAME_STATUS |
3267                                       DIFF_FORMAT_CHECKDIFF |
3268                                       DIFF_FORMAT_NO_OUTPUT))
3269                 options->output_format &= ~(DIFF_FORMAT_RAW |
3270                                             DIFF_FORMAT_NUMSTAT |
3271                                             DIFF_FORMAT_DIFFSTAT |
3272                                             DIFF_FORMAT_SHORTSTAT |
3273                                             DIFF_FORMAT_DIRSTAT |
3274                                             DIFF_FORMAT_SUMMARY |
3275                                             DIFF_FORMAT_PATCH);
3276
3277         /*
3278          * These cases always need recursive; we do not drop caller-supplied
3279          * recursive bits for other formats here.
3280          */
3281         if (options->output_format & (DIFF_FORMAT_PATCH |
3282                                       DIFF_FORMAT_NUMSTAT |
3283                                       DIFF_FORMAT_DIFFSTAT |
3284                                       DIFF_FORMAT_SHORTSTAT |
3285                                       DIFF_FORMAT_DIRSTAT |
3286                                       DIFF_FORMAT_SUMMARY |
3287                                       DIFF_FORMAT_CHECKDIFF))
3288                 DIFF_OPT_SET(options, RECURSIVE);
3289         /*
3290          * Also pickaxe would not work very well if you do not say recursive
3291          */
3292         if (options->pickaxe)
3293                 DIFF_OPT_SET(options, RECURSIVE);
3294         /*
3295          * When patches are generated, submodules diffed against the work tree
3296          * must be checked for dirtiness too so it can be shown in the output
3297          */
3298         if (options->output_format & DIFF_FORMAT_PATCH)
3299                 DIFF_OPT_SET(options, DIRTY_SUBMODULES);
3300
3301         if (options->detect_rename && options->rename_limit < 0)
3302                 options->rename_limit = diff_rename_limit_default;
3303         if (options->setup & DIFF_SETUP_USE_CACHE) {
3304                 if (!active_cache)
3305                         /* read-cache does not die even when it fails
3306                          * so it is safe for us to do this here.  Also
3307                          * it does not smudge active_cache or active_nr
3308                          * when it fails, so we do not have to worry about
3309                          * cleaning it up ourselves either.
3310                          */
3311                         read_cache();
3312         }
3313         if (options->abbrev <= 0 || 40 < options->abbrev)
3314                 options->abbrev = 40; /* full */
3315
3316         /*
3317          * It does not make sense to show the first hit we happened
3318          * to have found.  It does not make sense not to return with
3319          * exit code in such a case either.
3320          */
3321         if (DIFF_OPT_TST(options, QUICK)) {
3322                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
3323                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
3324         }
3325
3326         options->diff_path_counter = 0;
3327
3328         if (DIFF_OPT_TST(options, FOLLOW_RENAMES) && options->pathspec.nr != 1)
3329                 die(_("--follow requires exactly one pathspec"));
3330 }
3331
3332 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
3333 {
3334         char c, *eq;
3335         int len;
3336
3337         if (*arg != '-')
3338                 return 0;
3339         c = *++arg;
3340         if (!c)
3341                 return 0;
3342         if (c == arg_short) {
3343                 c = *++arg;
3344                 if (!c)
3345                         return 1;
3346                 if (val && isdigit(c)) {
3347                         char *end;
3348                         int n = strtoul(arg, &end, 10);
3349                         if (*end)
3350                                 return 0;
3351                         *val = n;
3352                         return 1;
3353                 }
3354                 return 0;
3355         }
3356         if (c != '-')
3357                 return 0;
3358         arg++;
3359         eq = strchrnul(arg, '=');
3360         len = eq - arg;
3361         if (!len || strncmp(arg, arg_long, len))
3362                 return 0;
3363         if (*eq) {
3364                 int n;
3365                 char *end;
3366                 if (!isdigit(*++eq))
3367                         return 0;
3368                 n = strtoul(eq, &end, 10);
3369                 if (*end)
3370                         return 0;
3371                 *val = n;
3372         }
3373         return 1;
3374 }
3375
3376 static int diff_scoreopt_parse(const char *opt);
3377
3378 static inline int short_opt(char opt, const char **argv,
3379                             const char **optarg)
3380 {
3381         const char *arg = argv[0];
3382         if (arg[0] != '-' || arg[1] != opt)
3383                 return 0;
3384         if (arg[2] != '\0') {
3385                 *optarg = arg + 2;
3386                 return 1;
3387         }
3388         if (!argv[1])
3389                 die("Option '%c' requires a value", opt);
3390         *optarg = argv[1];
3391         return 2;
3392 }
3393
3394 int parse_long_opt(const char *opt, const char **argv,
3395                    const char **optarg)
3396 {
3397         const char *arg = argv[0];
3398         if (!skip_prefix(arg, "--", &arg))
3399                 return 0;
3400         if (!skip_prefix(arg, opt, &arg))
3401                 return 0;
3402         if (*arg == '=') { /* stuck form: --option=value */
3403                 *optarg = arg + 1;
3404                 return 1;
3405         }
3406         if (*arg != '\0')
3407                 return 0;
3408         /* separate form: --option value */
3409         if (!argv[1])
3410                 die("Option '--%s' requires a value", opt);
3411         *optarg = argv[1];
3412         return 2;
3413 }
3414
3415 static int stat_opt(struct diff_options *options, const char **av)
3416 {
3417         const char *arg = av[0];
3418         char *end;
3419         int width = options->stat_width;
3420         int name_width = options->stat_name_width;
3421         int graph_width = options->stat_graph_width;
3422         int count = options->stat_count;
3423         int argcount = 1;
3424
3425         if (!skip_prefix(arg, "--stat", &arg))
3426                 die("BUG: stat option does not begin with --stat: %s", arg);
3427         end = (char *)arg;
3428
3429         switch (*arg) {
3430         case '-':
3431                 if (skip_prefix(arg, "-width", &arg)) {
3432                         if (*arg == '=')
3433                                 width = strtoul(arg + 1, &end, 10);
3434                         else if (!*arg && !av[1])
3435                                 die("Option '--stat-width' requires a value");
3436                         else if (!*arg) {
3437                                 width = strtoul(av[1], &end, 10);
3438                                 argcount = 2;
3439                         }
3440                 } else if (skip_prefix(arg, "-name-width", &arg)) {
3441                         if (*arg == '=')
3442                                 name_width = strtoul(arg + 1, &end, 10);
3443                         else if (!*arg && !av[1])
3444                                 die("Option '--stat-name-width' requires a value");
3445                         else if (!*arg) {
3446                                 name_width = strtoul(av[1], &end, 10);
3447                                 argcount = 2;
3448                         }
3449                 } else if (skip_prefix(arg, "-graph-width", &arg)) {
3450                         if (*arg == '=')
3451                                 graph_width = strtoul(arg + 1, &end, 10);
3452                         else if (!*arg && !av[1])
3453                                 die("Option '--stat-graph-width' requires a value");
3454                         else if (!*arg) {
3455                                 graph_width = strtoul(av[1], &end, 10);
3456                                 argcount = 2;
3457                         }
3458                 } else if (skip_prefix(arg, "-count", &arg)) {
3459                         if (*arg == '=')
3460                                 count = strtoul(arg + 1, &end, 10);
3461                         else if (!*arg && !av[1])
3462                                 die("Option '--stat-count' requires a value");
3463                         else if (!*arg) {
3464                                 count = strtoul(av[1], &end, 10);
3465                                 argcount = 2;
3466                         }
3467                 }
3468                 break;
3469         case '=':
3470                 width = strtoul(arg+1, &end, 10);
3471                 if (*end == ',')
3472                         name_width = strtoul(end+1, &end, 10);
3473                 if (*end == ',')
3474                         count = strtoul(end+1, &end, 10);
3475         }
3476
3477         /* Important! This checks all the error cases! */
3478         if (*end)
3479                 return 0;
3480         options->output_format |= DIFF_FORMAT_DIFFSTAT;
3481         options->stat_name_width = name_width;
3482         options->stat_graph_width = graph_width;
3483         options->stat_width = width;
3484         options->stat_count = count;
3485         return argcount;
3486 }
3487
3488 static int parse_dirstat_opt(struct diff_options *options, const char *params)
3489 {
3490         struct strbuf errmsg = STRBUF_INIT;
3491         if (parse_dirstat_params(options, params, &errmsg))
3492                 die(_("Failed to parse --dirstat/-X option parameter:\n%s"),
3493                     errmsg.buf);
3494         strbuf_release(&errmsg);
3495         /*
3496          * The caller knows a dirstat-related option is given from the command
3497          * line; allow it to say "return this_function();"
3498          */
3499         options->output_format |= DIFF_FORMAT_DIRSTAT;
3500         return 1;
3501 }
3502
3503 static int parse_submodule_opt(struct diff_options *options, const char *value)
3504 {
3505         if (parse_submodule_params(options, value))
3506                 die(_("Failed to parse --submodule option parameter: '%s'"),
3507                         value);
3508         return 1;
3509 }
3510
3511 static const char diff_status_letters[] = {
3512         DIFF_STATUS_ADDED,
3513         DIFF_STATUS_COPIED,
3514         DIFF_STATUS_DELETED,
3515         DIFF_STATUS_MODIFIED,
3516         DIFF_STATUS_RENAMED,
3517         DIFF_STATUS_TYPE_CHANGED,
3518         DIFF_STATUS_UNKNOWN,
3519         DIFF_STATUS_UNMERGED,
3520         DIFF_STATUS_FILTER_AON,
3521         DIFF_STATUS_FILTER_BROKEN,
3522         '\0',
3523 };
3524
3525 static unsigned int filter_bit['Z' + 1];
3526
3527 static void prepare_filter_bits(void)
3528 {
3529         int i;
3530
3531         if (!filter_bit[DIFF_STATUS_ADDED]) {
3532                 for (i = 0; diff_status_letters[i]; i++)
3533                         filter_bit[(int) diff_status_letters[i]] = (1 << i);
3534         }
3535 }
3536
3537 static unsigned filter_bit_tst(char status, const struct diff_options *opt)
3538 {
3539         return opt->filter & filter_bit[(int) status];
3540 }
3541
3542 static int parse_diff_filter_opt(const char *optarg, struct diff_options *opt)
3543 {
3544         int i, optch;
3545
3546         prepare_filter_bits();
3547
3548         /*
3549          * If there is a negation e.g. 'd' in the input, and we haven't
3550          * initialized the filter field with another --diff-filter, start
3551          * from full set of bits, except for AON.
3552          */
3553         if (!opt->filter) {
3554                 for (i = 0; (optch = optarg[i]) != '\0'; i++) {
3555                         if (optch < 'a' || 'z' < optch)
3556                                 continue;
3557                         opt->filter = (1 << (ARRAY_SIZE(diff_status_letters) - 1)) - 1;
3558                         opt->filter &= ~filter_bit[DIFF_STATUS_FILTER_AON];
3559                         break;
3560                 }
3561         }
3562
3563         for (i = 0; (optch = optarg[i]) != '\0'; i++) {
3564                 unsigned int bit;
3565                 int negate;
3566
3567                 if ('a' <= optch && optch <= 'z') {
3568                         negate = 1;
3569                         optch = toupper(optch);
3570                 } else {
3571                         negate = 0;
3572                 }
3573
3574                 bit = (0 <= optch && optch <= 'Z') ? filter_bit[optch] : 0;
3575                 if (!bit)
3576                         return optarg[i];
3577                 if (negate)
3578                         opt->filter &= ~bit;
3579                 else
3580                         opt->filter |= bit;
3581         }
3582         return 0;
3583 }
3584
3585 static void enable_patch_output(int *fmt) {
3586         *fmt &= ~DIFF_FORMAT_NO_OUTPUT;
3587         *fmt |= DIFF_FORMAT_PATCH;
3588 }
3589
3590 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
3591 {
3592         const char *arg = av[0];
3593         const char *optarg;
3594         int argcount;
3595
3596         /* Output format options */
3597         if (!strcmp(arg, "-p") || !strcmp(arg, "-u") || !strcmp(arg, "--patch")
3598             || opt_arg(arg, 'U', "unified", &options->context))
3599                 enable_patch_output(&options->output_format);
3600         else if (!strcmp(arg, "--raw"))
3601                 options->output_format |= DIFF_FORMAT_RAW;
3602         else if (!strcmp(arg, "--patch-with-raw")) {
3603                 enable_patch_output(&options->output_format);
3604                 options->output_format |= DIFF_FORMAT_RAW;
3605         } else if (!strcmp(arg, "--numstat"))
3606                 options->output_format |= DIFF_FORMAT_NUMSTAT;
3607         else if (!strcmp(arg, "--shortstat"))
3608                 options->output_format |= DIFF_FORMAT_SHORTSTAT;
3609         else if (!strcmp(arg, "-X") || !strcmp(arg, "--dirstat"))
3610                 return parse_dirstat_opt(options, "");
3611         else if (skip_prefix(arg, "-X", &arg))
3612                 return parse_dirstat_opt(options, arg);
3613         else if (skip_prefix(arg, "--dirstat=", &arg))
3614                 return parse_dirstat_opt(options, arg);
3615         else if (!strcmp(arg, "--cumulative"))
3616                 return parse_dirstat_opt(options, "cumulative");
3617         else if (!strcmp(arg, "--dirstat-by-file"))
3618                 return parse_dirstat_opt(options, "files");
3619         else if (skip_prefix(arg, "--dirstat-by-file=", &arg)) {
3620                 parse_dirstat_opt(options, "files");
3621                 return parse_dirstat_opt(options, arg);
3622         }
3623         else if (!strcmp(arg, "--check"))
3624                 options->output_format |= DIFF_FORMAT_CHECKDIFF;
3625         else if (!strcmp(arg, "--summary"))
3626                 options->output_format |= DIFF_FORMAT_SUMMARY;
3627         else if (!strcmp(arg, "--patch-with-stat")) {
3628                 enable_patch_output(&options->output_format);
3629                 options->output_format |= DIFF_FORMAT_DIFFSTAT;
3630         } else if (!strcmp(arg, "--name-only"))
3631                 options->output_format |= DIFF_FORMAT_NAME;
3632         else if (!strcmp(arg, "--name-status"))
3633                 options->output_format |= DIFF_FORMAT_NAME_STATUS;
3634         else if (!strcmp(arg, "-s") || !strcmp(arg, "--no-patch"))
3635                 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
3636         else if (starts_with(arg, "--stat"))
3637                 /* --stat, --stat-width, --stat-name-width, or --stat-count */
3638                 return stat_opt(options, av);
3639
3640         /* renames options */
3641         else if (starts_with(arg, "-B") || starts_with(arg, "--break-rewrites=") ||
3642                  !strcmp(arg, "--break-rewrites")) {
3643                 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
3644                         return error("invalid argument to -B: %s", arg+2);
3645         }
3646         else if (starts_with(arg, "-M") || starts_with(arg, "--find-renames=") ||
3647                  !strcmp(arg, "--find-renames")) {
3648                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
3649                         return error("invalid argument to -M: %s", arg+2);
3650                 options->detect_rename = DIFF_DETECT_RENAME;
3651         }
3652         else if (!strcmp(arg, "-D") || !strcmp(arg, "--irreversible-delete")) {
3653                 options->irreversible_delete = 1;
3654         }
3655         else if (starts_with(arg, "-C") || starts_with(arg, "--find-copies=") ||
3656                  !strcmp(arg, "--find-copies")) {
3657                 if (options->detect_rename == DIFF_DETECT_COPY)
3658                         DIFF_OPT_SET(options, FIND_COPIES_HARDER);
3659                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
3660                         return error("invalid argument to -C: %s", arg+2);
3661                 options->detect_rename = DIFF_DETECT_COPY;
3662         }
3663         else if (!strcmp(arg, "--no-renames"))
3664                 options->detect_rename = 0;
3665         else if (!strcmp(arg, "--rename-empty"))
3666                 DIFF_OPT_SET(options, RENAME_EMPTY);
3667         else if (!strcmp(arg, "--no-rename-empty"))
3668                 DIFF_OPT_CLR(options, RENAME_EMPTY);
3669         else if (!strcmp(arg, "--relative"))
3670                 DIFF_OPT_SET(options, RELATIVE_NAME);
3671         else if (skip_prefix(arg, "--relative=", &arg)) {
3672                 DIFF_OPT_SET(options, RELATIVE_NAME);
3673                 options->prefix = arg;
3674         }
3675
3676         /* xdiff options */
3677         else if (!strcmp(arg, "--minimal"))
3678                 DIFF_XDL_SET(options, NEED_MINIMAL);
3679         else if (!strcmp(arg, "--no-minimal"))
3680                 DIFF_XDL_CLR(options, NEED_MINIMAL);
3681         else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
3682                 DIFF_XDL_SET(options, IGNORE_WHITESPACE);
3683         else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
3684                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_CHANGE);
3685         else if (!strcmp(arg, "--ignore-space-at-eol"))
3686                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_AT_EOL);
3687         else if (!strcmp(arg, "--ignore-blank-lines"))
3688                 DIFF_XDL_SET(options, IGNORE_BLANK_LINES);
3689         else if (!strcmp(arg, "--patience"))
3690                 options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
3691         else if (!strcmp(arg, "--histogram"))
3692                 options->xdl_opts = DIFF_WITH_ALG(options, HISTOGRAM_DIFF);
3693         else if ((argcount = parse_long_opt("diff-algorithm", av, &optarg))) {
3694                 long value = parse_algorithm_value(optarg);
3695                 if (value < 0)
3696                         return error("option diff-algorithm accepts \"myers\", "
3697                                      "\"minimal\", \"patience\" and \"histogram\"");
3698                 /* clear out previous settings */
3699                 DIFF_XDL_CLR(options, NEED_MINIMAL);
3700                 options->xdl_opts &= ~XDF_DIFF_ALGORITHM_MASK;
3701                 options->xdl_opts |= value;
3702                 return argcount;
3703         }
3704
3705         /* flags options */
3706         else if (!strcmp(arg, "--binary")) {
3707                 enable_patch_output(&options->output_format);
3708                 DIFF_OPT_SET(options, BINARY);
3709         }
3710         else if (!strcmp(arg, "--full-index"))
3711                 DIFF_OPT_SET(options, FULL_INDEX);
3712         else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
3713                 DIFF_OPT_SET(options, TEXT);
3714         else if (!strcmp(arg, "-R"))
3715                 DIFF_OPT_SET(options, REVERSE_DIFF);
3716         else if (!strcmp(arg, "--find-copies-harder"))
3717                 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
3718         else if (!strcmp(arg, "--follow"))
3719                 DIFF_OPT_SET(options, FOLLOW_RENAMES);
3720         else if (!strcmp(arg, "--no-follow"))
3721                 DIFF_OPT_CLR(options, FOLLOW_RENAMES);
3722         else if (!strcmp(arg, "--color"))
3723                 options->use_color = 1;
3724         else if (skip_prefix(arg, "--color=", &arg)) {
3725                 int value = git_config_colorbool(NULL, arg);
3726                 if (value < 0)
3727                         return error("option `color' expects \"always\", \"auto\", or \"never\"");
3728                 options->use_color = value;
3729         }
3730         else if (!strcmp(arg, "--no-color"))
3731                 options->use_color = 0;
3732         else if (!strcmp(arg, "--color-words")) {
3733                 options->use_color = 1;
3734                 options->word_diff = DIFF_WORDS_COLOR;
3735         }
3736         else if (skip_prefix(arg, "--color-words=", &arg)) {
3737                 options->use_color = 1;
3738                 options->word_diff = DIFF_WORDS_COLOR;
3739                 options->word_regex = arg;
3740         }
3741         else if (!strcmp(arg, "--word-diff")) {
3742                 if (options->word_diff == DIFF_WORDS_NONE)
3743                         options->word_diff = DIFF_WORDS_PLAIN;
3744         }
3745         else if (skip_prefix(arg, "--word-diff=", &arg)) {
3746                 if (!strcmp(arg, "plain"))
3747                         options->word_diff = DIFF_WORDS_PLAIN;
3748                 else if (!strcmp(arg, "color")) {
3749                         options->use_color = 1;
3750                         options->word_diff = DIFF_WORDS_COLOR;
3751                 }
3752                 else if (!strcmp(arg, "porcelain"))
3753                         options->word_diff = DIFF_WORDS_PORCELAIN;
3754                 else if (!strcmp(arg, "none"))
3755                         options->word_diff = DIFF_WORDS_NONE;
3756                 else
3757                         die("bad --word-diff argument: %s", arg);
3758         }
3759         else if ((argcount = parse_long_opt("word-diff-regex", av, &optarg))) {
3760                 if (options->word_diff == DIFF_WORDS_NONE)
3761                         options->word_diff = DIFF_WORDS_PLAIN;
3762                 options->word_regex = optarg;
3763                 return argcount;
3764         }
3765         else if (!strcmp(arg, "--exit-code"))
3766                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
3767         else if (!strcmp(arg, "--quiet"))
3768                 DIFF_OPT_SET(options, QUICK);
3769         else if (!strcmp(arg, "--ext-diff"))
3770                 DIFF_OPT_SET(options, ALLOW_EXTERNAL);
3771         else if (!strcmp(arg, "--no-ext-diff"))
3772                 DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
3773         else if (!strcmp(arg, "--textconv"))
3774                 DIFF_OPT_SET(options, ALLOW_TEXTCONV);
3775         else if (!strcmp(arg, "--no-textconv"))
3776                 DIFF_OPT_CLR(options, ALLOW_TEXTCONV);
3777         else if (!strcmp(arg, "--ignore-submodules")) {
3778                 DIFF_OPT_SET(options, OVERRIDE_SUBMODULE_CONFIG);
3779                 handle_ignore_submodules_arg(options, "all");
3780         } else if (skip_prefix(arg, "--ignore-submodules=", &arg)) {
3781                 DIFF_OPT_SET(options, OVERRIDE_SUBMODULE_CONFIG);
3782                 handle_ignore_submodules_arg(options, arg);
3783         } else if (!strcmp(arg, "--submodule"))
3784                 DIFF_OPT_SET(options, SUBMODULE_LOG);
3785         else if (skip_prefix(arg, "--submodule=", &arg))
3786                 return parse_submodule_opt(options, arg);
3787
3788         /* misc options */
3789         else if (!strcmp(arg, "-z"))
3790                 options->line_termination = 0;
3791         else if ((argcount = short_opt('l', av, &optarg))) {
3792                 options->rename_limit = strtoul(optarg, NULL, 10);
3793                 return argcount;
3794         }
3795         else if ((argcount = short_opt('S', av, &optarg))) {
3796                 options->pickaxe = optarg;
3797                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_S;
3798                 return argcount;
3799         } else if ((argcount = short_opt('G', av, &optarg))) {
3800                 options->pickaxe = optarg;
3801                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_G;
3802                 return argcount;
3803         }
3804         else if (!strcmp(arg, "--pickaxe-all"))
3805                 options->pickaxe_opts |= DIFF_PICKAXE_ALL;
3806         else if (!strcmp(arg, "--pickaxe-regex"))
3807                 options->pickaxe_opts |= DIFF_PICKAXE_REGEX;
3808         else if ((argcount = short_opt('O', av, &optarg))) {
3809                 options->orderfile = optarg;
3810                 return argcount;
3811         }
3812         else if ((argcount = parse_long_opt("diff-filter", av, &optarg))) {
3813                 int offending = parse_diff_filter_opt(optarg, options);
3814                 if (offending)
3815                         die("unknown change class '%c' in --diff-filter=%s",
3816                             offending, optarg);
3817                 return argcount;
3818         }
3819         else if (!strcmp(arg, "--abbrev"))
3820                 options->abbrev = DEFAULT_ABBREV;
3821         else if (skip_prefix(arg, "--abbrev=", &arg)) {
3822                 options->abbrev = strtoul(arg, NULL, 10);
3823                 if (options->abbrev < MINIMUM_ABBREV)
3824                         options->abbrev = MINIMUM_ABBREV;
3825                 else if (40 < options->abbrev)
3826                         options->abbrev = 40;
3827         }
3828         else if ((argcount = parse_long_opt("src-prefix", av, &optarg))) {
3829                 options->a_prefix = optarg;
3830                 return argcount;
3831         }
3832         else if ((argcount = parse_long_opt("dst-prefix", av, &optarg))) {
3833                 options->b_prefix = optarg;
3834                 return argcount;
3835         }
3836         else if (!strcmp(arg, "--no-prefix"))
3837                 options->a_prefix = options->b_prefix = "";
3838         else if (opt_arg(arg, '\0', "inter-hunk-context",
3839                          &options->interhunkcontext))
3840                 ;
3841         else if (!strcmp(arg, "-W"))
3842                 DIFF_OPT_SET(options, FUNCCONTEXT);
3843         else if (!strcmp(arg, "--function-context"))
3844                 DIFF_OPT_SET(options, FUNCCONTEXT);
3845         else if (!strcmp(arg, "--no-function-context"))
3846                 DIFF_OPT_CLR(options, FUNCCONTEXT);
3847         else if ((argcount = parse_long_opt("output", av, &optarg))) {
3848                 options->file = fopen(optarg, "w");
3849                 if (!options->file)
3850                         die_errno("Could not open '%s'", optarg);
3851                 options->close_file = 1;
3852                 return argcount;
3853         } else
3854                 return 0;
3855         return 1;
3856 }
3857
3858 int parse_rename_score(const char **cp_p)
3859 {
3860         unsigned long num, scale;
3861         int ch, dot;
3862         const char *cp = *cp_p;
3863
3864         num = 0;
3865         scale = 1;
3866         dot = 0;
3867         for (;;) {
3868                 ch = *cp;
3869                 if ( !dot && ch == '.' ) {
3870                         scale = 1;
3871                         dot = 1;
3872                 } else if ( ch == '%' ) {
3873                         scale = dot ? scale*100 : 100;
3874                         cp++;   /* % is always at the end */
3875                         break;
3876                 } else if ( ch >= '0' && ch <= '9' ) {
3877                         if ( scale < 100000 ) {
3878                                 scale *= 10;
3879                                 num = (num*10) + (ch-'0');
3880                         }
3881                 } else {
3882                         break;
3883                 }
3884                 cp++;
3885         }
3886         *cp_p = cp;
3887
3888         /* user says num divided by scale and we say internally that
3889          * is MAX_SCORE * num / scale.
3890          */
3891         return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
3892 }
3893
3894 static int diff_scoreopt_parse(const char *opt)
3895 {
3896         int opt1, opt2, cmd;
3897
3898         if (*opt++ != '-')
3899                 return -1;
3900         cmd = *opt++;
3901         if (cmd == '-') {
3902                 /* convert the long-form arguments into short-form versions */
3903                 if (skip_prefix(opt, "break-rewrites", &opt)) {
3904                         if (*opt == 0 || *opt++ == '=')
3905                                 cmd = 'B';
3906                 } else if (skip_prefix(opt, "find-copies", &opt)) {
3907                         if (*opt == 0 || *opt++ == '=')
3908                                 cmd = 'C';
3909                 } else if (skip_prefix(opt, "find-renames", &opt)) {
3910                         if (*opt == 0 || *opt++ == '=')
3911                                 cmd = 'M';
3912                 }
3913         }
3914         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
3915                 return -1; /* that is not a -M, -C, or -B option */
3916
3917         opt1 = parse_rename_score(&opt);
3918         if (cmd != 'B')
3919                 opt2 = 0;
3920         else {
3921                 if (*opt == 0)
3922                         opt2 = 0;
3923                 else if (*opt != '/')
3924                         return -1; /* we expect -B80/99 or -B80 */
3925                 else {
3926                         opt++;
3927                         opt2 = parse_rename_score(&opt);
3928                 }
3929         }
3930         if (*opt != 0)
3931                 return -1;
3932         return opt1 | (opt2 << 16);
3933 }
3934
3935 struct diff_queue_struct diff_queued_diff;
3936
3937 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
3938 {
3939         ALLOC_GROW(queue->queue, queue->nr + 1, queue->alloc);
3940         queue->queue[queue->nr++] = dp;
3941 }
3942
3943 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
3944                                  struct diff_filespec *one,
3945                                  struct diff_filespec *two)
3946 {
3947         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
3948         dp->one = one;
3949         dp->two = two;
3950         if (queue)
3951                 diff_q(queue, dp);
3952         return dp;
3953 }
3954
3955 void diff_free_filepair(struct diff_filepair *p)
3956 {
3957         free_filespec(p->one);
3958         free_filespec(p->two);
3959         free(p);
3960 }
3961
3962 /* This is different from find_unique_abbrev() in that
3963  * it stuffs the result with dots for alignment.
3964  */
3965 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
3966 {
3967         int abblen;
3968         const char *abbrev;
3969         if (len == 40)
3970                 return sha1_to_hex(sha1);
3971
3972         abbrev = find_unique_abbrev(sha1, len);
3973         abblen = strlen(abbrev);
3974         if (abblen < 37) {
3975                 static char hex[41];
3976                 if (len < abblen && abblen <= len + 2)
3977                         sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
3978                 else
3979                         sprintf(hex, "%s...", abbrev);
3980                 return hex;
3981         }
3982         return sha1_to_hex(sha1);
3983 }
3984
3985 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
3986 {
3987         int line_termination = opt->line_termination;
3988         int inter_name_termination = line_termination ? '\t' : '\0';
3989
3990         fprintf(opt->file, "%s", diff_line_prefix(opt));
3991         if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
3992                 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
3993                         diff_unique_abbrev(p->one->sha1, opt->abbrev));
3994                 fprintf(opt->file, "%s ", diff_unique_abbrev(p->two->sha1, opt->abbrev));
3995         }
3996         if (p->score) {
3997                 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
3998                         inter_name_termination);
3999         } else {
4000                 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
4001         }
4002
4003         if (p->status == DIFF_STATUS_COPIED ||
4004             p->status == DIFF_STATUS_RENAMED) {
4005                 const char *name_a, *name_b;
4006                 name_a = p->one->path;
4007                 name_b = p->two->path;
4008                 strip_prefix(opt->prefix_length, &name_a, &name_b);
4009                 write_name_quoted(name_a, opt->file, inter_name_termination);
4010                 write_name_quoted(name_b, opt->file, line_termination);
4011         } else {
4012                 const char *name_a, *name_b;
4013                 name_a = p->one->mode ? p->one->path : p->two->path;
4014                 name_b = NULL;
4015                 strip_prefix(opt->prefix_length, &name_a, &name_b);
4016                 write_name_quoted(name_a, opt->file, line_termination);
4017         }
4018 }
4019
4020 int diff_unmodified_pair(struct diff_filepair *p)
4021 {
4022         /* This function is written stricter than necessary to support
4023          * the currently implemented transformers, but the idea is to
4024          * let transformers to produce diff_filepairs any way they want,
4025          * and filter and clean them up here before producing the output.
4026          */
4027         struct diff_filespec *one = p->one, *two = p->two;
4028
4029         if (DIFF_PAIR_UNMERGED(p))
4030                 return 0; /* unmerged is interesting */
4031
4032         /* deletion, addition, mode or type change
4033          * and rename are all interesting.
4034          */
4035         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
4036             DIFF_PAIR_MODE_CHANGED(p) ||
4037             strcmp(one->path, two->path))
4038                 return 0;
4039
4040         /* both are valid and point at the same path.  that is, we are
4041          * dealing with a change.
4042          */
4043         if (one->sha1_valid && two->sha1_valid &&
4044             !hashcmp(one->sha1, two->sha1) &&
4045             !one->dirty_submodule && !two->dirty_submodule)
4046                 return 1; /* no change */
4047         if (!one->sha1_valid && !two->sha1_valid)
4048                 return 1; /* both look at the same file on the filesystem. */
4049         return 0;
4050 }
4051
4052 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
4053 {
4054         if (diff_unmodified_pair(p))
4055                 return;
4056
4057         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4058             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4059                 return; /* no tree diffs in patch format */
4060
4061         run_diff(p, o);
4062 }
4063
4064 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
4065                             struct diffstat_t *diffstat)
4066 {
4067         if (diff_unmodified_pair(p))
4068                 return;
4069
4070         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4071             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4072                 return; /* no useful stat for tree diffs */
4073
4074         run_diffstat(p, o, diffstat);
4075 }
4076
4077 static void diff_flush_checkdiff(struct diff_filepair *p,
4078                 struct diff_options *o)
4079 {
4080         if (diff_unmodified_pair(p))
4081                 return;
4082
4083         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4084             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4085                 return; /* nothing to check in tree diffs */
4086
4087         run_checkdiff(p, o);
4088 }
4089
4090 int diff_queue_is_empty(void)
4091 {
4092         struct diff_queue_struct *q = &diff_queued_diff;
4093         int i;
4094         for (i = 0; i < q->nr; i++)
4095                 if (!diff_unmodified_pair(q->queue[i]))
4096                         return 0;
4097         return 1;
4098 }
4099
4100 #if DIFF_DEBUG
4101 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
4102 {
4103         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
4104                 x, one ? one : "",
4105                 s->path,
4106                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
4107                 s->mode,
4108                 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
4109         fprintf(stderr, "queue[%d] %s size %lu\n",
4110                 x, one ? one : "",
4111                 s->size);
4112 }
4113
4114 void diff_debug_filepair(const struct diff_filepair *p, int i)
4115 {
4116         diff_debug_filespec(p->one, i, "one");
4117         diff_debug_filespec(p->two, i, "two");
4118         fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
4119                 p->score, p->status ? p->status : '?',
4120                 p->one->rename_used, p->broken_pair);
4121 }
4122
4123 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
4124 {
4125         int i;
4126         if (msg)
4127                 fprintf(stderr, "%s\n", msg);
4128         fprintf(stderr, "q->nr = %d\n", q->nr);
4129         for (i = 0; i < q->nr; i++) {
4130                 struct diff_filepair *p = q->queue[i];
4131                 diff_debug_filepair(p, i);
4132         }
4133 }
4134 #endif
4135
4136 static void diff_resolve_rename_copy(void)
4137 {
4138         int i;
4139         struct diff_filepair *p;
4140         struct diff_queue_struct *q = &diff_queued_diff;
4141
4142         diff_debug_queue("resolve-rename-copy", q);
4143
4144         for (i = 0; i < q->nr; i++) {
4145                 p = q->queue[i];
4146                 p->status = 0; /* undecided */
4147                 if (DIFF_PAIR_UNMERGED(p))
4148                         p->status = DIFF_STATUS_UNMERGED;
4149                 else if (!DIFF_FILE_VALID(p->one))
4150                         p->status = DIFF_STATUS_ADDED;
4151                 else if (!DIFF_FILE_VALID(p->two))
4152                         p->status = DIFF_STATUS_DELETED;
4153                 else if (DIFF_PAIR_TYPE_CHANGED(p))
4154                         p->status = DIFF_STATUS_TYPE_CHANGED;
4155
4156                 /* from this point on, we are dealing with a pair
4157                  * whose both sides are valid and of the same type, i.e.
4158                  * either in-place edit or rename/copy edit.
4159                  */
4160                 else if (DIFF_PAIR_RENAME(p)) {
4161                         /*
4162                          * A rename might have re-connected a broken
4163                          * pair up, causing the pathnames to be the
4164                          * same again. If so, that's not a rename at
4165                          * all, just a modification..
4166                          *
4167                          * Otherwise, see if this source was used for
4168                          * multiple renames, in which case we decrement
4169                          * the count, and call it a copy.
4170                          */
4171                         if (!strcmp(p->one->path, p->two->path))
4172                                 p->status = DIFF_STATUS_MODIFIED;
4173                         else if (--p->one->rename_used > 0)
4174                                 p->status = DIFF_STATUS_COPIED;
4175                         else
4176                                 p->status = DIFF_STATUS_RENAMED;
4177                 }
4178                 else if (hashcmp(p->one->sha1, p->two->sha1) ||
4179                          p->one->mode != p->two->mode ||
4180                          p->one->dirty_submodule ||
4181                          p->two->dirty_submodule ||
4182                          is_null_sha1(p->one->sha1))
4183                         p->status = DIFF_STATUS_MODIFIED;
4184                 else {
4185                         /* This is a "no-change" entry and should not
4186                          * happen anymore, but prepare for broken callers.
4187                          */
4188                         error("feeding unmodified %s to diffcore",
4189                               p->one->path);
4190                         p->status = DIFF_STATUS_UNKNOWN;
4191                 }
4192         }
4193         diff_debug_queue("resolve-rename-copy done", q);
4194 }
4195
4196 static int check_pair_status(struct diff_filepair *p)
4197 {
4198         switch (p->status) {
4199         case DIFF_STATUS_UNKNOWN:
4200                 return 0;
4201         case 0:
4202                 die("internal error in diff-resolve-rename-copy");
4203         default:
4204                 return 1;
4205         }
4206 }
4207
4208 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
4209 {
4210         int fmt = opt->output_format;
4211
4212         if (fmt & DIFF_FORMAT_CHECKDIFF)
4213                 diff_flush_checkdiff(p, opt);
4214         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
4215                 diff_flush_raw(p, opt);
4216         else if (fmt & DIFF_FORMAT_NAME) {
4217                 const char *name_a, *name_b;
4218                 name_a = p->two->path;
4219                 name_b = NULL;
4220                 strip_prefix(opt->prefix_length, &name_a, &name_b);
4221                 write_name_quoted(name_a, opt->file, opt->line_termination);
4222         }
4223 }
4224
4225 static void show_file_mode_name(FILE *file, const char *newdelete, struct diff_filespec *fs)
4226 {
4227         if (fs->mode)
4228                 fprintf(file, " %s mode %06o ", newdelete, fs->mode);
4229         else
4230                 fprintf(file, " %s ", newdelete);
4231         write_name_quoted(fs->path, file, '\n');
4232 }
4233
4234
4235 static void show_mode_change(FILE *file, struct diff_filepair *p, int show_name,
4236                 const char *line_prefix)
4237 {
4238         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
4239                 fprintf(file, "%s mode change %06o => %06o%c", line_prefix, p->one->mode,
4240                         p->two->mode, show_name ? ' ' : '\n');
4241                 if (show_name) {
4242                         write_name_quoted(p->two->path, file, '\n');
4243                 }
4244         }
4245 }
4246
4247 static void show_rename_copy(FILE *file, const char *renamecopy, struct diff_filepair *p,
4248                         const char *line_prefix)
4249 {
4250         char *names = pprint_rename(p->one->path, p->two->path);
4251
4252         fprintf(file, " %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
4253         free(names);
4254         show_mode_change(file, p, 0, line_prefix);
4255 }
4256
4257 static void diff_summary(struct diff_options *opt, struct diff_filepair *p)
4258 {
4259         FILE *file = opt->file;
4260         const char *line_prefix = diff_line_prefix(opt);
4261
4262         switch(p->status) {
4263         case DIFF_STATUS_DELETED:
4264                 fputs(line_prefix, file);
4265                 show_file_mode_name(file, "delete", p->one);
4266                 break;
4267         case DIFF_STATUS_ADDED:
4268                 fputs(line_prefix, file);
4269                 show_file_mode_name(file, "create", p->two);
4270                 break;
4271         case DIFF_STATUS_COPIED:
4272                 fputs(line_prefix, file);
4273                 show_rename_copy(file, "copy", p, line_prefix);
4274                 break;
4275         case DIFF_STATUS_RENAMED:
4276                 fputs(line_prefix, file);
4277                 show_rename_copy(file, "rename", p, line_prefix);
4278                 break;
4279         default:
4280                 if (p->score) {
4281                         fprintf(file, "%s rewrite ", line_prefix);
4282                         write_name_quoted(p->two->path, file, ' ');
4283                         fprintf(file, "(%d%%)\n", similarity_index(p));
4284                 }
4285                 show_mode_change(file, p, !p->score, line_prefix);
4286                 break;
4287         }
4288 }
4289
4290 struct patch_id_t {
4291         git_SHA_CTX *ctx;
4292         int patchlen;
4293 };
4294
4295 static int remove_space(char *line, int len)
4296 {
4297         int i;
4298         char *dst = line;
4299         unsigned char c;
4300
4301         for (i = 0; i < len; i++)
4302                 if (!isspace((c = line[i])))
4303                         *dst++ = c;
4304
4305         return dst - line;
4306 }
4307
4308 static void patch_id_consume(void *priv, char *line, unsigned long len)
4309 {
4310         struct patch_id_t *data = priv;
4311         int new_len;
4312
4313         /* Ignore line numbers when computing the SHA1 of the patch */
4314         if (starts_with(line, "@@ -"))
4315                 return;
4316
4317         new_len = remove_space(line, len);
4318
4319         git_SHA1_Update(data->ctx, line, new_len);
4320         data->patchlen += new_len;
4321 }
4322
4323 /* returns 0 upon success, and writes result into sha1 */
4324 static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
4325 {
4326         struct diff_queue_struct *q = &diff_queued_diff;
4327         int i;
4328         git_SHA_CTX ctx;
4329         struct patch_id_t data;
4330         char buffer[PATH_MAX * 4 + 20];
4331
4332         git_SHA1_Init(&ctx);
4333         memset(&data, 0, sizeof(struct patch_id_t));
4334         data.ctx = &ctx;
4335
4336         for (i = 0; i < q->nr; i++) {
4337                 xpparam_t xpp;
4338                 xdemitconf_t xecfg;
4339                 mmfile_t mf1, mf2;
4340                 struct diff_filepair *p = q->queue[i];
4341                 int len1, len2;
4342
4343                 memset(&xpp, 0, sizeof(xpp));
4344                 memset(&xecfg, 0, sizeof(xecfg));
4345                 if (p->status == 0)
4346                         return error("internal diff status error");
4347                 if (p->status == DIFF_STATUS_UNKNOWN)
4348                         continue;
4349                 if (diff_unmodified_pair(p))
4350                         continue;
4351                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4352                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4353                         continue;
4354                 if (DIFF_PAIR_UNMERGED(p))
4355                         continue;
4356
4357                 diff_fill_sha1_info(p->one);
4358                 diff_fill_sha1_info(p->two);
4359                 if (fill_mmfile(&mf1, p->one) < 0 ||
4360                                 fill_mmfile(&mf2, p->two) < 0)
4361                         return error("unable to read files to diff");
4362
4363                 len1 = remove_space(p->one->path, strlen(p->one->path));
4364                 len2 = remove_space(p->two->path, strlen(p->two->path));
4365                 if (p->one->mode == 0)
4366                         len1 = snprintf(buffer, sizeof(buffer),
4367                                         "diff--gita/%.*sb/%.*s"
4368                                         "newfilemode%06o"
4369                                         "---/dev/null"
4370                                         "+++b/%.*s",
4371                                         len1, p->one->path,
4372                                         len2, p->two->path,
4373                                         p->two->mode,
4374                                         len2, p->two->path);
4375                 else if (p->two->mode == 0)
4376                         len1 = snprintf(buffer, sizeof(buffer),
4377                                         "diff--gita/%.*sb/%.*s"
4378                                         "deletedfilemode%06o"
4379                                         "---a/%.*s"
4380                                         "+++/dev/null",
4381                                         len1, p->one->path,
4382                                         len2, p->two->path,
4383                                         p->one->mode,
4384                                         len1, p->one->path);
4385                 else
4386                         len1 = snprintf(buffer, sizeof(buffer),
4387                                         "diff--gita/%.*sb/%.*s"
4388                                         "---a/%.*s"
4389                                         "+++b/%.*s",
4390                                         len1, p->one->path,
4391                                         len2, p->two->path,
4392                                         len1, p->one->path,
4393                                         len2, p->two->path);
4394                 git_SHA1_Update(&ctx, buffer, len1);
4395
4396                 if (diff_filespec_is_binary(p->one) ||
4397                     diff_filespec_is_binary(p->two)) {
4398                         git_SHA1_Update(&ctx, sha1_to_hex(p->one->sha1), 40);
4399                         git_SHA1_Update(&ctx, sha1_to_hex(p->two->sha1), 40);
4400                         continue;
4401                 }
4402
4403                 xpp.flags = 0;
4404                 xecfg.ctxlen = 3;
4405                 xecfg.flags = 0;
4406                 xdi_diff_outf(&mf1, &mf2, patch_id_consume, &data,
4407                               &xpp, &xecfg);
4408         }
4409
4410         git_SHA1_Final(sha1, &ctx);
4411         return 0;
4412 }
4413
4414 int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
4415 {
4416         struct diff_queue_struct *q = &diff_queued_diff;
4417         int i;
4418         int result = diff_get_patch_id(options, sha1);
4419
4420         for (i = 0; i < q->nr; i++)
4421                 diff_free_filepair(q->queue[i]);
4422
4423         free(q->queue);
4424         DIFF_QUEUE_CLEAR(q);
4425
4426         return result;
4427 }
4428
4429 static int is_summary_empty(const struct diff_queue_struct *q)
4430 {
4431         int i;
4432
4433         for (i = 0; i < q->nr; i++) {
4434                 const struct diff_filepair *p = q->queue[i];
4435
4436                 switch (p->status) {
4437                 case DIFF_STATUS_DELETED:
4438                 case DIFF_STATUS_ADDED:
4439                 case DIFF_STATUS_COPIED:
4440                 case DIFF_STATUS_RENAMED:
4441                         return 0;
4442                 default:
4443                         if (p->score)
4444                                 return 0;
4445                         if (p->one->mode && p->two->mode &&
4446                             p->one->mode != p->two->mode)
4447                                 return 0;
4448                         break;
4449                 }
4450         }
4451         return 1;
4452 }
4453
4454 static const char rename_limit_warning[] =
4455 "inexact rename detection was skipped due to too many files.";
4456
4457 static const char degrade_cc_to_c_warning[] =
4458 "only found copies from modified paths due to too many files.";
4459
4460 static const char rename_limit_advice[] =
4461 "you may want to set your %s variable to at least "
4462 "%d and retry the command.";
4463
4464 void diff_warn_rename_limit(const char *varname, int needed, int degraded_cc)
4465 {
4466         if (degraded_cc)
4467                 warning(degrade_cc_to_c_warning);
4468         else if (needed)
4469                 warning(rename_limit_warning);
4470         else
4471                 return;
4472         if (0 < needed && needed < 32767)
4473                 warning(rename_limit_advice, varname, needed);
4474 }
4475
4476 void diff_flush(struct diff_options *options)
4477 {
4478         struct diff_queue_struct *q = &diff_queued_diff;
4479         int i, output_format = options->output_format;
4480         int separator = 0;
4481         int dirstat_by_line = 0;
4482
4483         /*
4484          * Order: raw, stat, summary, patch
4485          * or:    name/name-status/checkdiff (other bits clear)
4486          */
4487         if (!q->nr)
4488                 goto free_queue;
4489
4490         if (output_format & (DIFF_FORMAT_RAW |
4491                              DIFF_FORMAT_NAME |
4492                              DIFF_FORMAT_NAME_STATUS |
4493                              DIFF_FORMAT_CHECKDIFF)) {
4494                 for (i = 0; i < q->nr; i++) {
4495                         struct diff_filepair *p = q->queue[i];
4496                         if (check_pair_status(p))
4497                                 flush_one_pair(p, options);
4498                 }
4499                 separator++;
4500         }
4501
4502         if (output_format & DIFF_FORMAT_DIRSTAT && DIFF_OPT_TST(options, DIRSTAT_BY_LINE))
4503                 dirstat_by_line = 1;
4504
4505         if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT) ||
4506             dirstat_by_line) {
4507                 struct diffstat_t diffstat;
4508
4509                 memset(&diffstat, 0, sizeof(struct diffstat_t));
4510                 for (i = 0; i < q->nr; i++) {
4511                         struct diff_filepair *p = q->queue[i];
4512                         if (check_pair_status(p))
4513                                 diff_flush_stat(p, options, &diffstat);
4514                 }
4515                 if (output_format & DIFF_FORMAT_NUMSTAT)
4516                         show_numstat(&diffstat, options);
4517                 if (output_format & DIFF_FORMAT_DIFFSTAT)
4518                         show_stats(&diffstat, options);
4519                 if (output_format & DIFF_FORMAT_SHORTSTAT)
4520                         show_shortstats(&diffstat, options);
4521                 if (output_format & DIFF_FORMAT_DIRSTAT)
4522                         show_dirstat_by_line(&diffstat, options);
4523                 free_diffstat_info(&diffstat);
4524                 separator++;
4525         }
4526         if ((output_format & DIFF_FORMAT_DIRSTAT) && !dirstat_by_line)
4527                 show_dirstat(options);
4528
4529         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
4530                 for (i = 0; i < q->nr; i++) {
4531                         diff_summary(options, q->queue[i]);
4532                 }
4533                 separator++;
4534         }
4535
4536         if (output_format & DIFF_FORMAT_NO_OUTPUT &&
4537             DIFF_OPT_TST(options, EXIT_WITH_STATUS) &&
4538             DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
4539                 /*
4540                  * run diff_flush_patch for the exit status. setting
4541                  * options->file to /dev/null should be safe, because we
4542                  * aren't supposed to produce any output anyway.
4543                  */
4544                 if (options->close_file)
4545                         fclose(options->file);
4546                 options->file = fopen("/dev/null", "w");
4547                 if (!options->file)
4548                         die_errno("Could not open /dev/null");
4549                 options->close_file = 1;
4550                 for (i = 0; i < q->nr; i++) {
4551                         struct diff_filepair *p = q->queue[i];
4552                         if (check_pair_status(p))
4553                                 diff_flush_patch(p, options);
4554                         if (options->found_changes)
4555                                 break;
4556                 }
4557         }
4558
4559         if (output_format & DIFF_FORMAT_PATCH) {
4560                 if (separator) {
4561                         fprintf(options->file, "%s%c",
4562                                 diff_line_prefix(options),
4563                                 options->line_termination);
4564                         if (options->stat_sep) {
4565                                 /* attach patch instead of inline */
4566                                 fputs(options->stat_sep, options->file);
4567                         }
4568                 }
4569
4570                 for (i = 0; i < q->nr; i++) {
4571                         struct diff_filepair *p = q->queue[i];
4572                         if (check_pair_status(p))
4573                                 diff_flush_patch(p, options);
4574                 }
4575         }
4576
4577         if (output_format & DIFF_FORMAT_CALLBACK)
4578                 options->format_callback(q, options, options->format_callback_data);
4579
4580         for (i = 0; i < q->nr; i++)
4581                 diff_free_filepair(q->queue[i]);
4582 free_queue:
4583         free(q->queue);
4584         DIFF_QUEUE_CLEAR(q);
4585         if (options->close_file)
4586                 fclose(options->file);
4587
4588         /*
4589          * Report the content-level differences with HAS_CHANGES;
4590          * diff_addremove/diff_change does not set the bit when
4591          * DIFF_FROM_CONTENTS is in effect (e.g. with -w).
4592          */
4593         if (DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
4594                 if (options->found_changes)
4595                         DIFF_OPT_SET(options, HAS_CHANGES);
4596                 else
4597                         DIFF_OPT_CLR(options, HAS_CHANGES);
4598         }
4599 }
4600
4601 static int match_filter(const struct diff_options *options, const struct diff_filepair *p)
4602 {
4603         return (((p->status == DIFF_STATUS_MODIFIED) &&
4604                  ((p->score &&
4605                    filter_bit_tst(DIFF_STATUS_FILTER_BROKEN, options)) ||
4606                   (!p->score &&
4607                    filter_bit_tst(DIFF_STATUS_MODIFIED, options)))) ||
4608                 ((p->status != DIFF_STATUS_MODIFIED) &&
4609                  filter_bit_tst(p->status, options)));
4610 }
4611
4612 static void diffcore_apply_filter(struct diff_options *options)
4613 {
4614         int i;
4615         struct diff_queue_struct *q = &diff_queued_diff;
4616         struct diff_queue_struct outq;
4617
4618         DIFF_QUEUE_CLEAR(&outq);
4619
4620         if (!options->filter)
4621                 return;
4622
4623         if (filter_bit_tst(DIFF_STATUS_FILTER_AON, options)) {
4624                 int found;
4625                 for (i = found = 0; !found && i < q->nr; i++) {
4626                         if (match_filter(options, q->queue[i]))
4627                                 found++;
4628                 }
4629                 if (found)
4630                         return;
4631
4632                 /* otherwise we will clear the whole queue
4633                  * by copying the empty outq at the end of this
4634                  * function, but first clear the current entries
4635                  * in the queue.
4636                  */
4637                 for (i = 0; i < q->nr; i++)
4638                         diff_free_filepair(q->queue[i]);
4639         }
4640         else {
4641                 /* Only the matching ones */
4642                 for (i = 0; i < q->nr; i++) {
4643                         struct diff_filepair *p = q->queue[i];
4644                         if (match_filter(options, p))
4645                                 diff_q(&outq, p);
4646                         else
4647                                 diff_free_filepair(p);
4648                 }
4649         }
4650         free(q->queue);
4651         *q = outq;
4652 }
4653
4654 /* Check whether two filespecs with the same mode and size are identical */
4655 static int diff_filespec_is_identical(struct diff_filespec *one,
4656                                       struct diff_filespec *two)
4657 {
4658         if (S_ISGITLINK(one->mode))
4659                 return 0;
4660         if (diff_populate_filespec(one, 0))
4661                 return 0;
4662         if (diff_populate_filespec(two, 0))
4663                 return 0;
4664         return !memcmp(one->data, two->data, one->size);
4665 }
4666
4667 static int diff_filespec_check_stat_unmatch(struct diff_filepair *p)
4668 {
4669         if (p->done_skip_stat_unmatch)
4670                 return p->skip_stat_unmatch_result;
4671
4672         p->done_skip_stat_unmatch = 1;
4673         p->skip_stat_unmatch_result = 0;
4674         /*
4675          * 1. Entries that come from stat info dirtiness
4676          *    always have both sides (iow, not create/delete),
4677          *    one side of the object name is unknown, with
4678          *    the same mode and size.  Keep the ones that
4679          *    do not match these criteria.  They have real
4680          *    differences.
4681          *
4682          * 2. At this point, the file is known to be modified,
4683          *    with the same mode and size, and the object
4684          *    name of one side is unknown.  Need to inspect
4685          *    the identical contents.
4686          */
4687         if (!DIFF_FILE_VALID(p->one) || /* (1) */
4688             !DIFF_FILE_VALID(p->two) ||
4689             (p->one->sha1_valid && p->two->sha1_valid) ||
4690             (p->one->mode != p->two->mode) ||
4691             diff_populate_filespec(p->one, 1) ||
4692             diff_populate_filespec(p->two, 1) ||
4693             (p->one->size != p->two->size) ||
4694             !diff_filespec_is_identical(p->one, p->two)) /* (2) */
4695                 p->skip_stat_unmatch_result = 1;
4696         return p->skip_stat_unmatch_result;
4697 }
4698
4699 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
4700 {
4701         int i;
4702         struct diff_queue_struct *q = &diff_queued_diff;
4703         struct diff_queue_struct outq;
4704         DIFF_QUEUE_CLEAR(&outq);
4705
4706         for (i = 0; i < q->nr; i++) {
4707                 struct diff_filepair *p = q->queue[i];
4708
4709                 if (diff_filespec_check_stat_unmatch(p))
4710                         diff_q(&outq, p);
4711                 else {
4712                         /*
4713                          * The caller can subtract 1 from skip_stat_unmatch
4714                          * to determine how many paths were dirty only
4715                          * due to stat info mismatch.
4716                          */
4717                         if (!DIFF_OPT_TST(diffopt, NO_INDEX))
4718                                 diffopt->skip_stat_unmatch++;
4719                         diff_free_filepair(p);
4720                 }
4721         }
4722         free(q->queue);
4723         *q = outq;
4724 }
4725
4726 static int diffnamecmp(const void *a_, const void *b_)
4727 {
4728         const struct diff_filepair *a = *((const struct diff_filepair **)a_);
4729         const struct diff_filepair *b = *((const struct diff_filepair **)b_);
4730         const char *name_a, *name_b;
4731
4732         name_a = a->one ? a->one->path : a->two->path;
4733         name_b = b->one ? b->one->path : b->two->path;
4734         return strcmp(name_a, name_b);
4735 }
4736
4737 void diffcore_fix_diff_index(struct diff_options *options)
4738 {
4739         struct diff_queue_struct *q = &diff_queued_diff;
4740         qsort(q->queue, q->nr, sizeof(q->queue[0]), diffnamecmp);
4741 }
4742
4743 void diffcore_std(struct diff_options *options)
4744 {
4745         /* NOTE please keep the following in sync with diff_tree_combined() */
4746         if (options->skip_stat_unmatch)
4747                 diffcore_skip_stat_unmatch(options);
4748         if (!options->found_follow) {
4749                 /* See try_to_follow_renames() in tree-diff.c */
4750                 if (options->break_opt != -1)
4751                         diffcore_break(options->break_opt);
4752                 if (options->detect_rename)
4753                         diffcore_rename(options);
4754                 if (options->break_opt != -1)
4755                         diffcore_merge_broken();
4756         }
4757         if (options->pickaxe)
4758                 diffcore_pickaxe(options);
4759         if (options->orderfile)
4760                 diffcore_order(options->orderfile);
4761         if (!options->found_follow)
4762                 /* See try_to_follow_renames() in tree-diff.c */
4763                 diff_resolve_rename_copy();
4764         diffcore_apply_filter(options);
4765
4766         if (diff_queued_diff.nr && !DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
4767                 DIFF_OPT_SET(options, HAS_CHANGES);
4768         else
4769                 DIFF_OPT_CLR(options, HAS_CHANGES);
4770
4771         options->found_follow = 0;
4772 }
4773
4774 int diff_result_code(struct diff_options *opt, int status)
4775 {
4776         int result = 0;
4777
4778         diff_warn_rename_limit("diff.renameLimit",
4779                                opt->needed_rename_limit,
4780                                opt->degraded_cc_to_c);
4781         if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
4782             !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
4783                 return status;
4784         if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
4785             DIFF_OPT_TST(opt, HAS_CHANGES))
4786                 result |= 01;
4787         if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
4788             DIFF_OPT_TST(opt, CHECK_FAILED))
4789                 result |= 02;
4790         return result;
4791 }
4792
4793 int diff_can_quit_early(struct diff_options *opt)
4794 {
4795         return (DIFF_OPT_TST(opt, QUICK) &&
4796                 !opt->filter &&
4797                 DIFF_OPT_TST(opt, HAS_CHANGES));
4798 }
4799
4800 /*
4801  * Shall changes to this submodule be ignored?
4802  *
4803  * Submodule changes can be configured to be ignored separately for each path,
4804  * but that configuration can be overridden from the command line.
4805  */
4806 static int is_submodule_ignored(const char *path, struct diff_options *options)
4807 {
4808         int ignored = 0;
4809         unsigned orig_flags = options->flags;
4810         if (!DIFF_OPT_TST(options, OVERRIDE_SUBMODULE_CONFIG))
4811                 set_diffopt_flags_from_submodule_config(options, path);
4812         if (DIFF_OPT_TST(options, IGNORE_SUBMODULES))
4813                 ignored = 1;
4814         options->flags = orig_flags;
4815         return ignored;
4816 }
4817
4818 void diff_addremove(struct diff_options *options,
4819                     int addremove, unsigned mode,
4820                     const unsigned char *sha1,
4821                     int sha1_valid,
4822                     const char *concatpath, unsigned dirty_submodule)
4823 {
4824         struct diff_filespec *one, *two;
4825
4826         if (S_ISGITLINK(mode) && is_submodule_ignored(concatpath, options))
4827                 return;
4828
4829         /* This may look odd, but it is a preparation for
4830          * feeding "there are unchanged files which should
4831          * not produce diffs, but when you are doing copy
4832          * detection you would need them, so here they are"
4833          * entries to the diff-core.  They will be prefixed
4834          * with something like '=' or '*' (I haven't decided
4835          * which but should not make any difference).
4836          * Feeding the same new and old to diff_change()
4837          * also has the same effect.
4838          * Before the final output happens, they are pruned after
4839          * merged into rename/copy pairs as appropriate.
4840          */
4841         if (DIFF_OPT_TST(options, REVERSE_DIFF))
4842                 addremove = (addremove == '+' ? '-' :
4843                              addremove == '-' ? '+' : addremove);
4844
4845         if (options->prefix &&
4846             strncmp(concatpath, options->prefix, options->prefix_length))
4847                 return;
4848
4849         one = alloc_filespec(concatpath);
4850         two = alloc_filespec(concatpath);
4851
4852         if (addremove != '+')
4853                 fill_filespec(one, sha1, sha1_valid, mode);
4854         if (addremove != '-') {
4855                 fill_filespec(two, sha1, sha1_valid, mode);
4856                 two->dirty_submodule = dirty_submodule;
4857         }
4858
4859         diff_queue(&diff_queued_diff, one, two);
4860         if (!DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
4861                 DIFF_OPT_SET(options, HAS_CHANGES);
4862 }
4863
4864 void diff_change(struct diff_options *options,
4865                  unsigned old_mode, unsigned new_mode,
4866                  const unsigned char *old_sha1,
4867                  const unsigned char *new_sha1,
4868                  int old_sha1_valid, int new_sha1_valid,
4869                  const char *concatpath,
4870                  unsigned old_dirty_submodule, unsigned new_dirty_submodule)
4871 {
4872         struct diff_filespec *one, *two;
4873         struct diff_filepair *p;
4874
4875         if (S_ISGITLINK(old_mode) && S_ISGITLINK(new_mode) &&
4876             is_submodule_ignored(concatpath, options))
4877                 return;
4878
4879         if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
4880                 unsigned tmp;
4881                 const unsigned char *tmp_c;
4882                 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
4883                 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
4884                 tmp = old_sha1_valid; old_sha1_valid = new_sha1_valid;
4885                         new_sha1_valid = tmp;
4886                 tmp = old_dirty_submodule; old_dirty_submodule = new_dirty_submodule;
4887                         new_dirty_submodule = tmp;
4888         }
4889
4890         if (options->prefix &&
4891             strncmp(concatpath, options->prefix, options->prefix_length))
4892                 return;
4893
4894         one = alloc_filespec(concatpath);
4895         two = alloc_filespec(concatpath);
4896         fill_filespec(one, old_sha1, old_sha1_valid, old_mode);
4897         fill_filespec(two, new_sha1, new_sha1_valid, new_mode);
4898         one->dirty_submodule = old_dirty_submodule;
4899         two->dirty_submodule = new_dirty_submodule;
4900         p = diff_queue(&diff_queued_diff, one, two);
4901
4902         if (DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
4903                 return;
4904
4905         if (DIFF_OPT_TST(options, QUICK) && options->skip_stat_unmatch &&
4906             !diff_filespec_check_stat_unmatch(p))
4907                 return;
4908
4909         DIFF_OPT_SET(options, HAS_CHANGES);
4910 }
4911
4912 struct diff_filepair *diff_unmerge(struct diff_options *options, const char *path)
4913 {
4914         struct diff_filepair *pair;
4915         struct diff_filespec *one, *two;
4916
4917         if (options->prefix &&
4918             strncmp(path, options->prefix, options->prefix_length))
4919                 return NULL;
4920
4921         one = alloc_filespec(path);
4922         two = alloc_filespec(path);
4923         pair = diff_queue(&diff_queued_diff, one, two);
4924         pair->is_unmerged = 1;
4925         return pair;
4926 }
4927
4928 static char *run_textconv(const char *pgm, struct diff_filespec *spec,
4929                 size_t *outsize)
4930 {
4931         struct diff_tempfile *temp;
4932         const char *argv[3];
4933         const char **arg = argv;
4934         struct child_process child;
4935         struct strbuf buf = STRBUF_INIT;
4936         int err = 0;
4937
4938         temp = prepare_temp_file(spec->path, spec);
4939         *arg++ = pgm;
4940         *arg++ = temp->name;
4941         *arg = NULL;
4942
4943         memset(&child, 0, sizeof(child));
4944         child.use_shell = 1;
4945         child.argv = argv;
4946         child.out = -1;
4947         if (start_command(&child)) {
4948                 remove_tempfile();
4949                 return NULL;
4950         }
4951
4952         if (strbuf_read(&buf, child.out, 0) < 0)
4953                 err = error("error reading from textconv command '%s'", pgm);
4954         close(child.out);
4955
4956         if (finish_command(&child) || err) {
4957                 strbuf_release(&buf);
4958                 remove_tempfile();
4959                 return NULL;
4960         }
4961         remove_tempfile();
4962
4963         return strbuf_detach(&buf, outsize);
4964 }
4965
4966 size_t fill_textconv(struct userdiff_driver *driver,
4967                      struct diff_filespec *df,
4968                      char **outbuf)
4969 {
4970         size_t size;
4971
4972         if (!driver || !driver->textconv) {
4973                 if (!DIFF_FILE_VALID(df)) {
4974                         *outbuf = "";
4975                         return 0;
4976                 }
4977                 if (diff_populate_filespec(df, 0))
4978                         die("unable to read files to diff");
4979                 *outbuf = df->data;
4980                 return df->size;
4981         }
4982
4983         if (driver->textconv_cache && df->sha1_valid) {
4984                 *outbuf = notes_cache_get(driver->textconv_cache, df->sha1,
4985                                           &size);
4986                 if (*outbuf)
4987                         return size;
4988         }
4989
4990         *outbuf = run_textconv(driver->textconv, df, &size);
4991         if (!*outbuf)
4992                 die("unable to read files to diff");
4993
4994         if (driver->textconv_cache && df->sha1_valid) {
4995                 /* ignore errors, as we might be in a readonly repository */
4996                 notes_cache_put(driver->textconv_cache, df->sha1, *outbuf,
4997                                 size);
4998                 /*
4999                  * we could save up changes and flush them all at the end,
5000                  * but we would need an extra call after all diffing is done.
5001                  * Since generating a cache entry is the slow path anyway,
5002                  * this extra overhead probably isn't a big deal.
5003                  */
5004                 notes_cache_write(driver->textconv_cache);
5005         }
5006
5007         return size;
5008 }
5009
5010 void setup_diff_pager(struct diff_options *opt)
5011 {
5012         /*
5013          * If the user asked for our exit code, then either they want --quiet
5014          * or --exit-code. We should definitely not bother with a pager in the
5015          * former case, as we will generate no output. Since we still properly
5016          * report our exit code even when a pager is run, we _could_ run a
5017          * pager with --exit-code. But since we have not done so historically,
5018          * and because it is easy to find people oneline advising "git diff
5019          * --exit-code" in hooks and other scripts, we do not do so.
5020          */
5021         if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
5022             check_pager_config("diff") != 0)
5023                 setup_pager();
5024 }